?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
configure: fix tcc_lddir, cpu
[tinycc.git] / tccelf.c
blob23c7e9b7d79478c8524877aa7d9415c40d71e392
1 /*
2 * ELF file handling for TCC
3 *
4 * Copyright (c) 2001-2004 Fabrice Bellard
5 *
6 * This library is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2 of the License, or (at your option) any later version.
10 *
11 * This library is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
15 *
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with this library; if not, write to the Free Software
18 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
19 */
20
21 #include "tcc.h"
22
23 /* Define this to get some debug output during relocation processing. */
24 #undef DEBUG_RELOC
25
26 /* XXX: avoid static variable */
27 static int new_undef_sym = 0; /* Is there a new undefined sym since last new_undef_sym() */
28
29 ST_FUNC int put_elf_str(Section *s, const char *sym)
30 {
31 int offset, len;
32 char *ptr;
33
34 len = strlen(sym) + 1;
35 offset = s->data_offset;
36 ptr = section_ptr_add(s, len);
37 memcpy(ptr, sym, len);
38 return offset;
39 }
40
41 /* elf symbol hashing function */
42 static unsigned long elf_hash(const unsigned char *name)
43 {
44 unsigned long h = 0, g;
45
46 while (*name) {
47 h = (h << 4) + *name++;
48 g = h & 0xf0000000;
49 if (g)
50 h ^= g >> 24;
51 h &= ~g;
52 }
53 return h;
54 }
55
56 /* rebuild hash table of section s */
57 /* NOTE: we do factorize the hash table code to go faster */
58 static void rebuild_hash(Section *s, unsigned int nb_buckets)
59 {
60 ElfW(Sym) *sym;
61 int *ptr, *hash, nb_syms, sym_index, h;
62 unsigned char *strtab;
63
64 strtab = s->link->data;
65 nb_syms = s->data_offset / sizeof(ElfW(Sym));
66
67 s->hash->data_offset = 0;
68 ptr = section_ptr_add(s->hash, (2 + nb_buckets + nb_syms) * sizeof(int));
69 ptr[0] = nb_buckets;
70 ptr[1] = nb_syms;
71 ptr += 2;
72 hash = ptr;
73 memset(hash, 0, (nb_buckets + 1) * sizeof(int));
74 ptr += nb_buckets + 1;
75
76 sym = (ElfW(Sym) *)s->data + 1;
77 for(sym_index = 1; sym_index < nb_syms; sym_index++) {
78 if (ELFW(ST_BIND)(sym->st_info) != STB_LOCAL) {
79 h = elf_hash(strtab + sym->st_name) % nb_buckets;
80 *ptr = hash[h];
81 hash[h] = sym_index;
82 } else {
83 *ptr = 0;
84 }
85 ptr++;
86 sym++;
87 }
88 }
89
90 /* return the symbol number */
91 ST_FUNC int put_elf_sym(Section *s, addr_t value, unsigned long size,
92 int info, int other, int shndx, const char *name)
93 {
94 int name_offset, sym_index;
95 int nbuckets, h;
96 ElfW(Sym) *sym;
97 Section *hs;
98
99 sym = section_ptr_add(s, sizeof(ElfW(Sym)));
100 if (name)
101 name_offset = put_elf_str(s->link, name);
102 else
103 name_offset = 0;
104 /* XXX: endianness */
105 sym->st_name = name_offset;
106 sym->st_value = value;
107 sym->st_size = size;
108 sym->st_info = info;
109 sym->st_other = other;
110 sym->st_shndx = shndx;
111 sym_index = sym - (ElfW(Sym) *)s->data;
112 hs = s->hash;
113 if (hs) {
114 int *ptr, *base;
115 ptr = section_ptr_add(hs, sizeof(int));
116 base = (int *)hs->data;
117 /* only add global or weak symbols */
118 if (ELFW(ST_BIND)(info) != STB_LOCAL) {
119 /* add another hashing entry */
120 nbuckets = base[0];
121 h = elf_hash((unsigned char *) name) % nbuckets;
122 *ptr = base[2 + h];
123 base[2 + h] = sym_index;
124 base[1]++;
125 /* we resize the hash table */
126 hs->nb_hashed_syms++;
127 if (hs->nb_hashed_syms > 2 * nbuckets) {
128 rebuild_hash(s, 2 * nbuckets);
129 }
130 } else {
131 *ptr = 0;
132 base[1]++;
133 }
134 }
135 return sym_index;
136 }
137
138 /* find global ELF symbol 'name' and return its index. Return 0 if not
139 found. */
140 ST_FUNC int find_elf_sym(Section *s, const char *name)
141 {
142 ElfW(Sym) *sym;
143 Section *hs;
144 int nbuckets, sym_index, h;
145 const char *name1;
146
147 hs = s->hash;
148 if (!hs)
149 return 0;
150 nbuckets = ((int *)hs->data)[0];
151 h = elf_hash((unsigned char *) name) % nbuckets;
152 sym_index = ((int *)hs->data)[2 + h];
153 while (sym_index != 0) {
154 sym = &((ElfW(Sym) *)s->data)[sym_index];
155 name1 = (char *) s->link->data + sym->st_name;
156 if (!strcmp(name, name1))
157 return sym_index;
158 sym_index = ((int *)hs->data)[2 + nbuckets + sym_index];
159 }
160 return 0;
161 }
162
163 /* return elf symbol value, signal error if 'err' is nonzero */
164 ST_FUNC addr_t get_elf_sym_addr(TCCState *s, const char *name, int err)
165 {
166 int sym_index;
167 ElfW(Sym) *sym;
168
169 sym_index = find_elf_sym(s->symtab, name);
170 sym = &((ElfW(Sym) *)s->symtab->data)[sym_index];
171 if (!sym_index || sym->st_shndx == SHN_UNDEF) {
172 if (err)
173 tcc_error("%s not defined", name);
174 return 0;
175 }
176 return sym->st_value;
177 }
178
179 /* return elf symbol value */
180 LIBTCCAPI void *tcc_get_symbol(TCCState *s, const char *name)
181 {
182 return (void*)(uintptr_t)get_elf_sym_addr(s, name, 0);
183 }
184
185 #if defined TCC_IS_NATIVE || defined TCC_TARGET_PE
186 /* return elf symbol value or error */
187 ST_FUNC void* tcc_get_symbol_err(TCCState *s, const char *name)
188 {
189 return (void*)(uintptr_t)get_elf_sym_addr(s, name, 1);
190 }
191 #endif
192
193 /* add an elf symbol : check if it is already defined and patch
194 it. Return symbol index. NOTE that sh_num can be SHN_UNDEF. */
195 ST_FUNC int add_elf_sym(Section *s, addr_t value, unsigned long size,
196 int info, int other, int sh_num, const char *name)
197 {
198 ElfW(Sym) *esym;
199 int sym_bind, sym_index, sym_type, esym_bind;
200 unsigned char sym_vis, esym_vis, new_vis;
201
202 sym_bind = ELFW(ST_BIND)(info);
203 sym_type = ELFW(ST_TYPE)(info);
204 sym_vis = ELFW(ST_VISIBILITY)(other);
205
206 if (sym_bind != STB_LOCAL) {
207 /* we search global or weak symbols */
208 sym_index = find_elf_sym(s, name);
209 if (!sym_index)
210 goto do_def;
211 esym = &((ElfW(Sym) *)s->data)[sym_index];
212 if (esym->st_shndx != SHN_UNDEF) {
213 esym_bind = ELFW(ST_BIND)(esym->st_info);
214 /* propagate the most constraining visibility */
215 /* STV_DEFAULT(0)<STV_PROTECTED(3)<STV_HIDDEN(2)<STV_INTERNAL(1) */
216 esym_vis = ELFW(ST_VISIBILITY)(esym->st_other);
217 if (esym_vis == STV_DEFAULT) {
218 new_vis = sym_vis;
219 } else if (sym_vis == STV_DEFAULT) {
220 new_vis = esym_vis;
221 } else {
222 new_vis = (esym_vis < sym_vis) ? esym_vis : sym_vis;
223 }
224 esym->st_other = (esym->st_other & ~ELFW(ST_VISIBILITY)(-1))
225 | new_vis;
226 other = esym->st_other; /* in case we have to patch esym */
227 if (sh_num == SHN_UNDEF) {
228 /* ignore adding of undefined symbol if the
229 corresponding symbol is already defined */
230 } else if (sym_bind == STB_GLOBAL && esym_bind == STB_WEAK) {
231 /* global overrides weak, so patch */
232 goto do_patch;
233 } else if (sym_bind == STB_WEAK && esym_bind == STB_GLOBAL) {
234 /* weak is ignored if already global */
235 } else if (sym_bind == STB_WEAK && esym_bind == STB_WEAK) {
236 /* keep first-found weak definition, ignore subsequents */
237 } else if (sym_vis == STV_HIDDEN || sym_vis == STV_INTERNAL) {
238 /* ignore hidden symbols after */
239 } else if ((esym->st_shndx == SHN_COMMON
240 || esym->st_shndx == bss_section->sh_num)
241 && (sh_num < SHN_LORESERVE
242 && sh_num != bss_section->sh_num)) {
243 /* data symbol gets precedence over common/bss */
244 goto do_patch;
245 } else if (sh_num == SHN_COMMON || sh_num == bss_section->sh_num) {
246 /* data symbol keeps precedence over common/bss */
247 } else if (s == tcc_state->dynsymtab_section) {
248 /* we accept that two DLL define the same symbol */
249 } else {
250 #if 0
251 printf("new_bind=%x new_shndx=%x new_vis=%x old_bind=%x old_shndx=%x old_vis=%x\n",
252 sym_bind, sh_num, new_vis, esym_bind, esym->st_shndx, esym_vis);
253 #endif
254 tcc_error_noabort("'%s' defined twice", name);
255 }
256 } else {
257 do_patch:
258 esym->st_info = ELFW(ST_INFO)(sym_bind, sym_type);
259 esym->st_shndx = sh_num;
260 new_undef_sym = 1;
261 esym->st_value = value;
262 esym->st_size = size;
263 esym->st_other = other;
264 }
265 } else {
266 do_def:
267 sym_index = put_elf_sym(s, value, size,
268 ELFW(ST_INFO)(sym_bind, sym_type), other,
269 sh_num, name);
270 }
271 return sym_index;
272 }
273
274 /* put relocation */
275 ST_FUNC void put_elf_reloca(Section *symtab, Section *s, unsigned long offset,
276 int type, int symbol, addr_t addend)
277 {
278 char buf[256];
279 Section *sr;
280 ElfW_Rel *rel;
281
282 sr = s->reloc;
283 if (!sr) {
284 /* if no relocation section, create it */
285 snprintf(buf, sizeof(buf), REL_SECTION_FMT, s->name);
286 /* if the symtab is allocated, then we consider the relocation
287 are also */
288 sr = new_section(tcc_state, buf, SHT_RELX, symtab->sh_flags);
289 sr->sh_entsize = sizeof(ElfW_Rel);
290 sr->link = symtab;
291 sr->sh_info = s->sh_num;
292 s->reloc = sr;
293 }
294 rel = section_ptr_add(sr, sizeof(ElfW_Rel));
295 rel->r_offset = offset;
296 rel->r_info = ELFW(R_INFO)(symbol, type);
297 #if defined(TCC_TARGET_ARM64) || defined(TCC_TARGET_X86_64)
298 rel->r_addend = addend;
299 #else
300 if (addend)
301 tcc_error("non-zero addend on REL architecture");
302 #endif
303 }
304
305 ST_FUNC void put_elf_reloc(Section *symtab, Section *s, unsigned long offset,
306 int type, int symbol)
307 {
308 put_elf_reloca(symtab, s, offset, type, symbol, 0);
309 }
310
311 /* put stab debug information */
312
313 ST_FUNC void put_stabs(const char *str, int type, int other, int desc,
314 unsigned long value)
315 {
316 Stab_Sym *sym;
317
318 sym = section_ptr_add(stab_section, sizeof(Stab_Sym));
319 if (str) {
320 sym->n_strx = put_elf_str(stabstr_section, str);
321 } else {
322 sym->n_strx = 0;
323 }
324 sym->n_type = type;
325 sym->n_other = other;
326 sym->n_desc = desc;
327 sym->n_value = value;
328 }
329
330 ST_FUNC void put_stabs_r(const char *str, int type, int other, int desc,
331 unsigned long value, Section *sec, int sym_index)
332 {
333 put_stabs(str, type, other, desc, value);
334 put_elf_reloc(symtab_section, stab_section,
335 stab_section->data_offset - sizeof(unsigned int),
336 R_DATA_32, sym_index);
337 }
338
339 ST_FUNC void put_stabn(int type, int other, int desc, int value)
340 {
341 put_stabs(NULL, type, other, desc, value);
342 }
343
344 ST_FUNC void put_stabd(int type, int other, int desc)
345 {
346 put_stabs(NULL, type, other, desc, 0);
347 }
348
349 /* Browse each elem of type <type> in section <sec> starting at elem <startoff>
350 using variable <elem> */
351 #define for_each_elem(sec, startoff, elem, type) \
352 for (elem = (type *) sec->data + startoff; \
353 elem < (type *) (sec->data + sec->data_offset); elem++)
354
355 /* In an ELF file symbol table, the local symbols must appear below
356 the global and weak ones. Since TCC cannot sort it while generating
357 the code, we must do it after. All the relocation tables are also
358 modified to take into account the symbol table sorting */
359 static void sort_syms(TCCState *s1, Section *s)
360 {
361 int *old_to_new_syms;
362 ElfW(Sym) *new_syms;
363 int nb_syms, i;
364 ElfW(Sym) *p, *q;
365 ElfW_Rel *rel;
366 Section *sr;
367 int type, sym_index;
368
369 nb_syms = s->data_offset / sizeof(ElfW(Sym));
370 new_syms = tcc_malloc(nb_syms * sizeof(ElfW(Sym)));
371 old_to_new_syms = tcc_malloc(nb_syms * sizeof(int));
372
373 /* first pass for local symbols */
374 p = (ElfW(Sym) *)s->data;
375 q = new_syms;
376 for(i = 0; i < nb_syms; i++) {
377 if (ELFW(ST_BIND)(p->st_info) == STB_LOCAL) {
378 old_to_new_syms[i] = q - new_syms;
379 *q++ = *p;
380 }
381 p++;
382 }
383 /* save the number of local symbols in section header */
384 s->sh_info = q - new_syms;
385
386 /* then second pass for non local symbols */
387 p = (ElfW(Sym) *)s->data;
388 for(i = 0; i < nb_syms; i++) {
389 if (ELFW(ST_BIND)(p->st_info) != STB_LOCAL) {
390 old_to_new_syms[i] = q - new_syms;
391 *q++ = *p;
392 }
393 p++;
394 }
395
396 /* we copy the new symbols to the old */
397 memcpy(s->data, new_syms, nb_syms * sizeof(ElfW(Sym)));
398 tcc_free(new_syms);
399
400 /* now we modify all the relocations */
401 for(i = 1; i < s1->nb_sections; i++) {
402 sr = s1->sections[i];
403 if (sr->sh_type == SHT_RELX && sr->link == s) {
404 for_each_elem(sr, 0, rel, ElfW_Rel) {
405 sym_index = ELFW(R_SYM)(rel->r_info);
406 type = ELFW(R_TYPE)(rel->r_info);
407 sym_index = old_to_new_syms[sym_index];
408 rel->r_info = ELFW(R_INFO)(sym_index, type);
409 }
410 }
411 }
412
413 tcc_free(old_to_new_syms);
414 }
415
416 /* relocate common symbols in the .bss section */
417 ST_FUNC void relocate_common_syms(void)
418 {
419 ElfW(Sym) *sym;
420 unsigned long offset, align;
421
422 for_each_elem(symtab_section, 1, sym, ElfW(Sym)) {
423 if (sym->st_shndx == SHN_COMMON) {
424 /* align symbol */
425 align = sym->st_value;
426 offset = bss_section->data_offset;
427 offset = (offset + align - 1) & -align;
428 sym->st_value = offset;
429 sym->st_shndx = bss_section->sh_num;
430 offset += sym->st_size;
431 bss_section->data_offset = offset;
432 }
433 }
434 }
435
436 /* relocate symbol table, resolve undefined symbols if do_resolve is
437 true and output error if undefined symbol. */
438 ST_FUNC void relocate_syms(TCCState *s1, int do_resolve)
439 {
440 ElfW(Sym) *sym, *esym;
441 int sym_bind, sh_num, sym_index;
442 const char *name;
443
444 for_each_elem(symtab_section, 1, sym, ElfW(Sym)) {
445 sh_num = sym->st_shndx;
446 if (sh_num == SHN_UNDEF) {
447 name = (char *) strtab_section->data + sym->st_name;
448 /* Use ld.so to resolve symbol for us (for tcc -run) */
449 if (do_resolve) {
450 #if defined TCC_IS_NATIVE && !defined _WIN32
451 void *addr;
452 name = (char *) symtab_section->link->data + sym->st_name;
453 addr = resolve_sym(s1, name);
454 if (addr) {
455 sym->st_value = (addr_t)addr;
456 #ifdef DEBUG_RELOC
457 printf ("relocate_sym: %s -> 0x%lx\n", name, sym->st_value);
458 #endif
459 goto found;
460 }
461 #endif
462 } else if (s1->dynsym) {
463 /* if dynamic symbol exist, then use it */
464 sym_index = find_elf_sym(s1->dynsym, name);
465 if (sym_index) {
466 esym = &((ElfW(Sym) *)s1->dynsym->data)[sym_index];
467 sym->st_value = esym->st_value;
468 goto found;
469 }
470 }
471 /* XXX: _fp_hw seems to be part of the ABI, so we ignore
472 it */
473 if (!strcmp(name, "_fp_hw"))
474 goto found;
475 /* only weak symbols are accepted to be undefined. Their
476 value is zero */
477 sym_bind = ELFW(ST_BIND)(sym->st_info);
478 if (sym_bind == STB_WEAK) {
479 sym->st_value = 0;
480 } else {
481 tcc_error_noabort("undefined symbol '%s'", name);
482 }
483 } else if (sh_num < SHN_LORESERVE) {
484 /* add section base */
485 sym->st_value += s1->sections[sym->st_shndx]->sh_addr;
486 }
487 found: ;
488 }
489 }
490
491 /* relocate a given section (CPU dependent) by applying the relocations
492 in the associated relocation section */
493 ST_FUNC void relocate_section(TCCState *s1, Section *s)
494 {
495 Section *sr = s->reloc;
496 ElfW_Rel *rel;
497 ElfW(Sym) *sym;
498 int type, sym_index;
499 unsigned char *ptr;
500 addr_t val, addr;
501 #if defined TCC_TARGET_I386 || defined TCC_TARGET_X86_64
502 ElfW_Rel *qrel = (ElfW_Rel *) sr->data; /* ptr to next reloc entry reused */
503 int esym_index;
504 #endif
505
506 for_each_elem(sr, 0, rel, ElfW_Rel) {
507 ptr = s->data + rel->r_offset;
508
509 sym_index = ELFW(R_SYM)(rel->r_info);
510 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
511 val = sym->st_value;
512 #if defined(TCC_TARGET_ARM64) || defined(TCC_TARGET_X86_64)
513 val += rel->r_addend;
514 #endif
515 type = ELFW(R_TYPE)(rel->r_info);
516 addr = s->sh_addr + rel->r_offset;
517
518 /* CPU specific */
519 switch(type) {
520 #if defined(TCC_TARGET_I386)
521 case R_386_32:
522 if (s1->output_type == TCC_OUTPUT_DLL) {
523 esym_index = s1->symtab_to_dynsym[sym_index];
524 qrel->r_offset = rel->r_offset;
525 if (esym_index) {
526 qrel->r_info = ELFW(R_INFO)(esym_index, R_386_32);
527 qrel++;
528 break;
529 } else {
530 qrel->r_info = ELFW(R_INFO)(0, R_386_RELATIVE);
531 qrel++;
532 }
533 }
534 write32le(ptr, read32le(ptr) + val);
535 break;
536 case R_386_PC32:
537 if (s1->output_type == TCC_OUTPUT_DLL) {
538 /* DLL relocation */
539 esym_index = s1->symtab_to_dynsym[sym_index];
540 if (esym_index) {
541 qrel->r_offset = rel->r_offset;
542 qrel->r_info = ELFW(R_INFO)(esym_index, R_386_PC32);
543 qrel++;
544 break;
545 }
546 }
547 write32le(ptr, read32le(ptr) + val - addr);
548 break;
549 case R_386_PLT32:
550 write32le(ptr, read32le(ptr) + val - addr);
551 break;
552 case R_386_GLOB_DAT:
553 case R_386_JMP_SLOT:
554 write32le(ptr, val);
555 break;
556 case R_386_GOTPC:
557 write32le(ptr, read32le(ptr) + s1->got->sh_addr - addr);
558 break;
559 case R_386_GOTOFF:
560 write32le(ptr, read32le(ptr) + val - s1->got->sh_addr);
561 break;
562 case R_386_GOT32:
563 case R_386_GOT32X:
564 /* we load the got offset */
565 write32le(ptr, read32le(ptr) + s1->sym_attrs[sym_index].got_offset);
566 break;
567 case R_386_16:
568 if (s1->output_format != TCC_OUTPUT_FORMAT_BINARY) {
569 output_file:
570 tcc_error("can only produce 16-bit binary files");
571 }
572 write16le(ptr, read16le(ptr) + val);
573 break;
574 case R_386_PC16:
575 if (s1->output_format != TCC_OUTPUT_FORMAT_BINARY)
576 goto output_file;
577 write16le(ptr, read16le(ptr) + val - addr);
578 break;
579 case R_386_RELATIVE:
580 /* do nothing */
581 break;
582 case R_386_COPY:
583 /* This reloction must copy initialized data from the library
584 to the program .bss segment. Currently made like for ARM
585 (to remove noise of defaukt case). Is this true?
586 */
587 break;
588 default:
589 fprintf(stderr,"FIXME: handle reloc type %d at %x [%p] to %x\n",
590 type, (unsigned)addr, ptr, (unsigned)val);
591 break;
592 #elif defined(TCC_TARGET_ARM)
593 case R_ARM_PC24:
594 case R_ARM_CALL:
595 case R_ARM_JUMP24:
596 case R_ARM_PLT32:
597 {
598 int x, is_thumb, is_call, h, blx_avail, is_bl, th_ko;
599 x = (*(int *) ptr) & 0xffffff;
600 if (sym->st_shndx == SHN_UNDEF)
601 val = s1->plt->sh_addr;
602 #ifdef DEBUG_RELOC
603 printf ("reloc %d: x=0x%x val=0x%x ", type, x, val);
604 #endif
605 (*(int *)ptr) &= 0xff000000;
606 if (x & 0x800000)
607 x -= 0x1000000;
608 x <<= 2;
609 blx_avail = (TCC_ARM_VERSION >= 5);
610 is_thumb = val & 1;
611 is_bl = (*(unsigned *) ptr) >> 24 == 0xeb;
612 is_call = (type == R_ARM_CALL || (type == R_ARM_PC24 && is_bl));
613 x += val - addr;
614 #ifdef DEBUG_RELOC
615 printf (" newx=0x%x name=%s\n", x,
616 (char *) symtab_section->link->data + sym->st_name);
617 #endif
618 h = x & 2;
619 th_ko = (x & 3) && (!blx_avail || !is_call);
620 if (th_ko || x >= 0x2000000 || x < -0x2000000)
621 tcc_error("can't relocate value at %x,%d",addr, type);
622 x >>= 2;
623 x &= 0xffffff;
624 /* Only reached if blx is avail and it is a call */
625 if (is_thumb) {
626 x |= h << 24;
627 (*(int *)ptr) = 0xfa << 24; /* bl -> blx */
628 }
629 (*(int *) ptr) |= x;
630 }
631 break;
632 /* Since these relocations only concern Thumb-2 and blx instruction was
633 introduced before Thumb-2, we can assume blx is available and not
634 guard its use */
635 case R_ARM_THM_PC22:
636 case R_ARM_THM_JUMP24:
637 {
638 int x, hi, lo, s, j1, j2, i1, i2, imm10, imm11;
639 int to_thumb, is_call, to_plt, blx_bit = 1 << 12;
640 Section *plt;
641
642 /* weak reference */
643 if (sym->st_shndx == SHN_UNDEF &&
644 ELFW(ST_BIND)(sym->st_info) == STB_WEAK)
645 break;
646
647 /* Get initial offset */
648 hi = (*(uint16_t *)ptr);
649 lo = (*(uint16_t *)(ptr+2));
650 s = (hi >> 10) & 1;
651 j1 = (lo >> 13) & 1;
652 j2 = (lo >> 11) & 1;
653 i1 = (j1 ^ s) ^ 1;
654 i2 = (j2 ^ s) ^ 1;
655 imm10 = hi & 0x3ff;
656 imm11 = lo & 0x7ff;
657 x = (s << 24) | (i1 << 23) | (i2 << 22) |
658 (imm10 << 12) | (imm11 << 1);
659 if (x & 0x01000000)
660 x -= 0x02000000;
661
662 /* Relocation infos */
663 to_thumb = val & 1;
664 plt = s1->plt;
665 to_plt = (val >= plt->sh_addr) &&
666 (val < plt->sh_addr + plt->data_offset);
667 is_call = (type == R_ARM_THM_PC22);
668
669 /* Compute final offset */
670 if (to_plt && !is_call) /* Point to 1st instr of Thumb stub */
671 x -= 4;
672 x += val - addr;
673 if (!to_thumb && is_call) {
674 blx_bit = 0; /* bl -> blx */
675 x = (x + 3) & -4; /* Compute offset from aligned PC */
676 }
677
678 /* Check that relocation is possible
679 * offset must not be out of range
680 * if target is to be entered in arm mode:
681 - bit 1 must not set
682 - instruction must be a call (bl) or a jump to PLT */
683 if (!to_thumb || x >= 0x1000000 || x < -0x1000000)
684 if (to_thumb || (val & 2) || (!is_call && !to_plt))
685 tcc_error("can't relocate value at %x,%d",addr, type);
686
687 /* Compute and store final offset */
688 s = (x >> 24) & 1;
689 i1 = (x >> 23) & 1;
690 i2 = (x >> 22) & 1;
691 j1 = s ^ (i1 ^ 1);
692 j2 = s ^ (i2 ^ 1);
693 imm10 = (x >> 12) & 0x3ff;
694 imm11 = (x >> 1) & 0x7ff;
695 (*(uint16_t *)ptr) = (uint16_t) ((hi & 0xf800) |
696 (s << 10) | imm10);
697 (*(uint16_t *)(ptr+2)) = (uint16_t) ((lo & 0xc000) |
698 (j1 << 13) | blx_bit | (j2 << 11) |
699 imm11);
700 }
701 break;
702 case R_ARM_MOVT_ABS:
703 case R_ARM_MOVW_ABS_NC:
704 {
705 int x, imm4, imm12;
706 if (type == R_ARM_MOVT_ABS)
707 val >>= 16;
708 imm12 = val & 0xfff;
709 imm4 = (val >> 12) & 0xf;
710 x = (imm4 << 16) | imm12;
711 if (type == R_ARM_THM_MOVT_ABS)
712 *(int *)ptr |= x;
713 else
714 *(int *)ptr += x;
715 }
716 break;
717 case R_ARM_THM_MOVT_ABS:
718 case R_ARM_THM_MOVW_ABS_NC:
719 {
720 int x, i, imm4, imm3, imm8;
721 if (type == R_ARM_THM_MOVT_ABS)
722 val >>= 16;
723 imm8 = val & 0xff;
724 imm3 = (val >> 8) & 0x7;
725 i = (val >> 11) & 1;
726 imm4 = (val >> 12) & 0xf;
727 x = (imm3 << 28) | (imm8 << 16) | (i << 10) | imm4;
728 if (type == R_ARM_THM_MOVT_ABS)
729 *(int *)ptr |= x;
730 else
731 *(int *)ptr += x;
732 }
733 break;
734 case R_ARM_PREL31:
735 {
736 int x;
737 x = (*(int *)ptr) & 0x7fffffff;
738 (*(int *)ptr) &= 0x80000000;
739 x = (x * 2) / 2;
740 x += val - addr;
741 if((x^(x>>1))&0x40000000)
742 tcc_error("can't relocate value at %x,%d",addr, type);
743 (*(int *)ptr) |= x & 0x7fffffff;
744 }
745 case R_ARM_ABS32:
746 *(int *)ptr += val;
747 break;
748 case R_ARM_REL32:
749 *(int *)ptr += val - addr;
750 break;
751 case R_ARM_GOTPC:
752 *(int *)ptr += s1->got->sh_addr - addr;
753 break;
754 case R_ARM_GOTOFF:
755 *(int *)ptr += val - s1->got->sh_addr;
756 break;
757 case R_ARM_GOT32:
758 /* we load the got offset */
759 *(int *)ptr += s1->sym_attrs[sym_index].got_offset;
760 break;
761 case R_ARM_COPY:
762 break;
763 case R_ARM_V4BX:
764 /* trade Thumb support for ARMv4 support */
765 if ((0x0ffffff0 & *(int*)ptr) == 0x012FFF10)
766 *(int*)ptr ^= 0xE12FFF10 ^ 0xE1A0F000; /* BX Rm -> MOV PC, Rm */
767 break;
768 case R_ARM_GLOB_DAT:
769 case R_ARM_JUMP_SLOT:
770 *(addr_t *)ptr = val;
771 break;
772 case R_ARM_NONE:
773 /* Nothing to do. Normally used to indicate a dependency
774 on a certain symbol (like for exception handling under EABI). */
775 break;
776 default:
777 fprintf(stderr,"FIXME: handle reloc type %x at %x [%p] to %x\n",
778 type, (unsigned)addr, ptr, (unsigned)val);
779 break;
780 #elif defined(TCC_TARGET_ARM64)
781 case R_AARCH64_ABS64:
782 write64le(ptr, val);
783 break;
784 case R_AARCH64_ABS32:
785 write32le(ptr, val);
786 break;
787 case R_AARCH64_MOVW_UABS_G0_NC:
788 write32le(ptr, ((read32le(ptr) & 0xffe0001f) |
789 (val & 0xffff) << 5));
790 break;
791 case R_AARCH64_MOVW_UABS_G1_NC:
792 write32le(ptr, ((read32le(ptr) & 0xffe0001f) |
793 (val >> 16 & 0xffff) << 5));
794 break;
795 case R_AARCH64_MOVW_UABS_G2_NC:
796 write32le(ptr, ((read32le(ptr) & 0xffe0001f) |
797 (val >> 32 & 0xffff) << 5));
798 break;
799 case R_AARCH64_MOVW_UABS_G3:
800 write32le(ptr, ((read32le(ptr) & 0xffe0001f) |
801 (val >> 48 & 0xffff) << 5));
802 break;
803 case R_AARCH64_ADR_PREL_PG_HI21: {
804 uint64_t off = (val >> 12) - (addr >> 12);
805 if ((off + ((uint64_t)1 << 20)) >> 21)
806 tcc_error("R_AARCH64_ADR_PREL_PG_HI21 relocation failed");
807 write32le(ptr, ((read32le(ptr) & 0x9f00001f) |
808 (off & 0x1ffffc) << 3 | (off & 3) << 29));
809 break;
810 }
811 case R_AARCH64_ADD_ABS_LO12_NC:
812 write32le(ptr, ((read32le(ptr) & 0xffc003ff) |
813 (val & 0xfff) << 10));
814 break;
815 case R_AARCH64_JUMP26:
816 case R_AARCH64_CALL26:
817 /* This check must match the one in build_got_entries, testing
818 if we really need a PLT slot. */
819 if (sym->st_shndx == SHN_UNDEF)
820 /* We've put the PLT slot offset into r_addend when generating
821 it, and that's what we must use as relocation value (adjusted
822 by section offset of course). */
823 val = s1->plt->sh_addr + rel->r_addend;
824 #ifdef DEBUG_RELOC
825 printf ("reloc %d @ 0x%lx: val=0x%lx name=%s\n", type, addr, val,
826 (char *) symtab_section->link->data + sym->st_name);
827 #endif
828 if (((val - addr) + ((uint64_t)1 << 27)) & ~(uint64_t)0xffffffc)
829 {
830 tcc_error("R_AARCH64_(JUMP|CALL)26 relocation failed (val=%lx, addr=%lx)", addr, val);
831 }
832 write32le(ptr, (0x14000000 |
833 (uint32_t)(type == R_AARCH64_CALL26) << 31 |
834 ((val - addr) >> 2 & 0x3ffffff)));
835 break;
836 case R_AARCH64_ADR_GOT_PAGE: {
837 uint64_t off =
838 (((s1->got->sh_addr +
839 s1->sym_attrs[sym_index].got_offset) >> 12) - (addr >> 12));
840 if ((off + ((uint64_t)1 << 20)) >> 21)
841 tcc_error("R_AARCH64_ADR_GOT_PAGE relocation failed");
842 write32le(ptr, ((read32le(ptr) & 0x9f00001f) |
843 (off & 0x1ffffc) << 3 | (off & 3) << 29));
844 break;
845 }
846 case R_AARCH64_LD64_GOT_LO12_NC:
847 write32le(ptr,
848 ((read32le(ptr) & 0xfff803ff) |
849 ((s1->got->sh_addr +
850 s1->sym_attrs[sym_index].got_offset) & 0xff8) << 7));
851 break;
852 case R_AARCH64_COPY:
853 break;
854 case R_AARCH64_GLOB_DAT:
855 case R_AARCH64_JUMP_SLOT:
856 /* They don't need addend */
857 #ifdef DEBUG_RELOC
858 printf ("reloc %d @ 0x%lx: val=0x%lx name=%s\n", type, addr,
859 val - rel->r_addend,
860 (char *) symtab_section->link->data + sym->st_name);
861 #endif
862 write64le(ptr, val - rel->r_addend);
863 break;
864 default:
865 fprintf(stderr, "FIXME: handle reloc type %x at %x [%p] to %x\n",
866 type, (unsigned)addr, ptr, (unsigned)val);
867 break;
868 #elif defined(TCC_TARGET_C67)
869 case R_C60_32:
870 *(int *)ptr += val;
871 break;
872 case R_C60LO16:
873 {
874 uint32_t orig;
875
876 /* put the low 16 bits of the absolute address
877 add to what is already there */
878
879 orig = ((*(int *)(ptr )) >> 7) & 0xffff;
880 orig |= (((*(int *)(ptr+4)) >> 7) & 0xffff) << 16;
881
882 /* patch both at once - assumes always in pairs Low - High */
883
884 *(int *) ptr = (*(int *) ptr & (~(0xffff << 7)) ) | (((val+orig) & 0xffff) << 7);
885 *(int *)(ptr+4) = (*(int *)(ptr+4) & (~(0xffff << 7)) ) | ((((val+orig)>>16) & 0xffff) << 7);
886 }
887 break;
888 case R_C60HI16:
889 break;
890 default:
891 fprintf(stderr,"FIXME: handle reloc type %x at %x [%p] to %x\n",
892 type, (unsigned)addr, ptr, (unsigned)val);
893 break;
894 #elif defined(TCC_TARGET_X86_64)
895 case R_X86_64_64:
896 if (s1->output_type == TCC_OUTPUT_DLL) {
897 esym_index = s1->symtab_to_dynsym[sym_index];
898 qrel->r_offset = rel->r_offset;
899 if (esym_index) {
900 qrel->r_info = ELFW(R_INFO)(esym_index, R_X86_64_64);
901 qrel->r_addend = rel->r_addend;
902 qrel++;
903 break;
904 } else {
905 qrel->r_info = ELFW(R_INFO)(0, R_X86_64_RELATIVE);
906 qrel->r_addend = read64le(ptr) + val;
907 qrel++;
908 }
909 }
910 write64le(ptr, read64le(ptr) + val);
911 break;
912 case R_X86_64_32:
913 case R_X86_64_32S:
914 if (s1->output_type == TCC_OUTPUT_DLL) {
915 /* XXX: this logic may depend on TCC's codegen
916 now TCC uses R_X86_64_32 even for a 64bit pointer */
917 qrel->r_info = ELFW(R_INFO)(0, R_X86_64_RELATIVE);
918 /* Use sign extension! */
919 qrel->r_addend = (int)read32le(ptr) + val;
920 qrel++;
921 }
922 write32le(ptr, read32le(ptr) + val);
923 break;
924
925 case R_X86_64_PC32:
926 if (s1->output_type == TCC_OUTPUT_DLL) {
927 /* DLL relocation */
928 esym_index = s1->symtab_to_dynsym[sym_index];
929 if (esym_index) {
930 qrel->r_offset = rel->r_offset;
931 qrel->r_info = ELFW(R_INFO)(esym_index, R_X86_64_PC32);
932 /* Use sign extension! */
933 qrel->r_addend = (int)read32le(ptr) + rel->r_addend;
934 qrel++;
935 break;
936 }
937 }
938 goto plt32pc32;
939
940 case R_X86_64_PLT32:
941 /* We've put the PLT slot offset into r_addend when generating
942 it, and that's what we must use as relocation value (adjusted
943 by section offset of course). */
944 val = s1->plt->sh_addr + rel->r_addend;
945 /* fallthrough. */
946
947 plt32pc32:
948 {
949 long long diff;
950 diff = (long long)val - addr;
951 if (diff < -2147483648LL || diff > 2147483647LL) {
952 tcc_error("internal error: relocation failed");
953 }
954 write32le(ptr, read32le(ptr) + diff);
955 }
956 break;
957 case R_X86_64_GLOB_DAT:
958 case R_X86_64_JUMP_SLOT:
959 /* They don't need addend */
960 write64le(ptr, val - rel->r_addend);
961 break;
962 case R_X86_64_GOTPCREL:
963 case R_X86_64_GOTPCRELX:
964 case R_X86_64_REX_GOTPCRELX:
965 write32le(ptr, read32le(ptr) +
966 (s1->got->sh_addr - addr +
967 s1->sym_attrs[sym_index].got_offset - 4));
968 break;
969 case R_X86_64_GOTTPOFF:
970 write32le(ptr, read32le(ptr) + val - s1->got->sh_addr);
971 break;
972 case R_X86_64_GOT32:
973 /* we load the got offset */
974 write32le(ptr, read32le(ptr) + s1->sym_attrs[sym_index].got_offset);
975 break;
976 #else
977 #error unsupported processor
978 #endif
979 }
980 }
981 /* if the relocation is allocated, we change its symbol table */
982 if (sr->sh_flags & SHF_ALLOC)
983 sr->link = s1->dynsym;
984 }
985
986 /* relocate relocation table in 'sr' */
987 static void relocate_rel(TCCState *s1, Section *sr)
988 {
989 Section *s;
990 ElfW_Rel *rel;
991
992 s = s1->sections[sr->sh_info];
993 for_each_elem(sr, 0, rel, ElfW_Rel)
994 rel->r_offset += s->sh_addr;
995 }
996
997 /* count the number of dynamic relocations so that we can reserve
998 their space */
999 static int prepare_dynamic_rel(TCCState *s1, Section *sr)
1000 {
1001 ElfW_Rel *rel;
1002 int sym_index, esym_index, type, count;
1003
1004 count = 0;
1005 for_each_elem(sr, 0, rel, ElfW_Rel) {
1006 sym_index = ELFW(R_SYM)(rel->r_info);
1007 type = ELFW(R_TYPE)(rel->r_info);
1008 switch(type) {
1009 #if defined(TCC_TARGET_I386)
1010 case R_386_32:
1011 #elif defined(TCC_TARGET_X86_64)
1012 case R_X86_64_32:
1013 case R_X86_64_32S:
1014 case R_X86_64_64:
1015 #endif
1016 count++;
1017 break;
1018 #if defined(TCC_TARGET_I386)
1019 case R_386_PC32:
1020 #elif defined(TCC_TARGET_X86_64)
1021 case R_X86_64_PC32:
1022 #endif
1023 esym_index = s1->symtab_to_dynsym[sym_index];
1024 if (esym_index)
1025 count++;
1026 break;
1027 default:
1028 break;
1029 }
1030 }
1031 if (count) {
1032 /* allocate the section */
1033 sr->sh_flags |= SHF_ALLOC;
1034 sr->sh_size = count * sizeof(ElfW_Rel);
1035 }
1036 return count;
1037 }
1038
1039 static struct sym_attr *alloc_sym_attr(TCCState *s1, int index)
1040 {
1041 int n;
1042 struct sym_attr *tab;
1043
1044 if (index >= s1->nb_sym_attrs) {
1045 /* find immediately bigger power of 2 and reallocate array */
1046 n = 1;
1047 while (index >= n)
1048 n *= 2;
1049 tab = tcc_realloc(s1->sym_attrs, n * sizeof(*s1->sym_attrs));
1050 s1->sym_attrs = tab;
1051 memset(s1->sym_attrs + s1->nb_sym_attrs, 0,
1052 (n - s1->nb_sym_attrs) * sizeof(*s1->sym_attrs));
1053 s1->nb_sym_attrs = n;
1054 }
1055 return &s1->sym_attrs[index];
1056 }
1057
1058 static void build_got(TCCState *s1)
1059 {
1060 unsigned char *ptr;
1061
1062 /* if no got, then create it */
1063 s1->got = new_section(s1, ".got", SHT_PROGBITS, SHF_ALLOC | SHF_WRITE);
1064 s1->got->sh_entsize = 4;
1065 add_elf_sym(symtab_section, 0, 4, ELFW(ST_INFO)(STB_GLOBAL, STT_OBJECT),
1066 0, s1->got->sh_num, "_GLOBAL_OFFSET_TABLE_");
1067 ptr = section_ptr_add(s1->got, 3 * PTR_SIZE);
1068 #if PTR_SIZE == 4
1069 /* keep space for _DYNAMIC pointer, if present */
1070 write32le(ptr, 0);
1071 /* two dummy got entries */
1072 write32le(ptr + 4, 0);
1073 write32le(ptr + 8, 0);
1074 #else
1075 /* keep space for _DYNAMIC pointer, if present */
1076 write32le(ptr, 0);
1077 write32le(ptr + 4, 0);
1078 /* two dummy got entries */
1079 write32le(ptr + 8, 0);
1080 write32le(ptr + 12, 0);
1081 write32le(ptr + 16, 0);
1082 write32le(ptr + 20, 0);
1083 #endif
1084 }
1085
1086 /* put a got or plt entry corresponding to a symbol in symtab_section. 'size'
1087 and 'info' can be modifed if more precise info comes from the DLL.
1088 Returns offset of GOT or PLT slot. */
1089 static unsigned long put_got_entry(TCCState *s1,
1090 int reloc_type, unsigned long size, int info,
1091 int sym_index)
1092 {
1093 int index, need_plt_entry;
1094 const char *name;
1095 ElfW(Sym) *sym;
1096 unsigned long offset;
1097 int *ptr;
1098 struct sym_attr *symattr;
1099
1100 if (!s1->got)
1101 build_got(s1);
1102
1103 need_plt_entry =
1104 #ifdef TCC_TARGET_X86_64
1105 (reloc_type == R_X86_64_JUMP_SLOT);
1106 #elif defined(TCC_TARGET_I386)
1107 (reloc_type == R_386_JMP_SLOT);
1108 #elif defined(TCC_TARGET_ARM)
1109 (reloc_type == R_ARM_JUMP_SLOT);
1110 #elif defined(TCC_TARGET_ARM64)
1111 (reloc_type == R_AARCH64_JUMP_SLOT);
1112 #else
1113 0;
1114 #endif
1115
1116 if (need_plt_entry && !s1->plt) {
1117 /* add PLT */
1118 s1->plt = new_section(s1, ".plt", SHT_PROGBITS,
1119 SHF_ALLOC | SHF_EXECINSTR);
1120 s1->plt->sh_entsize = 4;
1121 }
1122
1123 /* If a got/plt entry already exists for that symbol, no need to add one */
1124 if (sym_index < s1->nb_sym_attrs) {
1125 if (need_plt_entry && s1->sym_attrs[sym_index].plt_offset)
1126 return s1->sym_attrs[sym_index].plt_offset;
1127 else if (!need_plt_entry && s1->sym_attrs[sym_index].got_offset)
1128 return s1->sym_attrs[sym_index].got_offset;
1129 }
1130
1131 symattr = alloc_sym_attr(s1, sym_index);
1132
1133 /* Only store the GOT offset if it's not generated for the PLT entry. */
1134 if (!need_plt_entry)
1135 symattr->got_offset = s1->got->data_offset;
1136
1137 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1138 name = (char *) symtab_section->link->data + sym->st_name;
1139 offset = sym->st_value;
1140 #if defined(TCC_TARGET_I386) || defined(TCC_TARGET_X86_64)
1141 if (need_plt_entry) {
1142 Section *plt;
1143 uint8_t *p;
1144 int modrm;
1145 unsigned long relofs;
1146
1147 #if defined(TCC_OUTPUT_DLL_WITH_PLT)
1148 modrm = 0x25;
1149 #else
1150 /* if we build a DLL, we add a %ebx offset */
1151 if (s1->output_type == TCC_OUTPUT_DLL)
1152 modrm = 0xa3;
1153 else
1154 modrm = 0x25;
1155 #endif
1156
1157 /* add a PLT entry */
1158 plt = s1->plt;
1159 if (plt->data_offset == 0) {
1160 /* first plt entry */
1161 p = section_ptr_add(plt, 16);
1162 p[0] = 0xff; /* pushl got + PTR_SIZE */
1163 p[1] = modrm + 0x10;
1164 write32le(p + 2, PTR_SIZE);
1165 p[6] = 0xff; /* jmp *(got + PTR_SIZE * 2) */
1166 p[7] = modrm;
1167 write32le(p + 8, PTR_SIZE * 2);
1168 }
1169
1170 /* The PLT slot refers to the relocation entry it needs
1171 via offset. The reloc entry is created below, so its
1172 offset is the current data_offset. */
1173 relofs = s1->got->reloc ? s1->got->reloc->data_offset : 0;
1174 symattr->plt_offset = plt->data_offset;
1175 p = section_ptr_add(plt, 16);
1176 p[0] = 0xff; /* jmp *(got + x) */
1177 p[1] = modrm;
1178 write32le(p + 2, s1->got->data_offset);
1179 p[6] = 0x68; /* push $xxx */
1180 #ifdef TCC_TARGET_X86_64
1181 /* On x86-64, the relocation is referred to by _index_. */
1182 write32le(p + 7, relofs / sizeof (ElfW_Rel));
1183 #else
1184 write32le(p + 7, relofs);
1185 #endif
1186 p[11] = 0xe9; /* jmp plt_start */
1187 write32le(p + 12, -(plt->data_offset));
1188
1189 /* If this was an UNDEF symbol set the offset in the
1190 dynsymtab to the PLT slot, so that PC32 relocs to it
1191 can be resolved. */
1192 if (sym->st_shndx == SHN_UNDEF)
1193 offset = plt->data_offset - 16;
1194 }
1195 #elif defined(TCC_TARGET_ARM)
1196 if (need_plt_entry) {
1197 Section *plt;
1198 uint8_t *p;
1199
1200 /* if we build a DLL, we add a %ebx offset */
1201 if (s1->output_type == TCC_OUTPUT_DLL)
1202 tcc_error("DLLs unimplemented!");
1203
1204 /* add a PLT entry */
1205 plt = s1->plt;
1206 if (plt->data_offset == 0) {
1207 /* first plt entry */
1208 p = section_ptr_add(plt, 16);
1209 write32le(p, 0xe52de004); /* push {lr} */
1210 write32le(p+4, 0xe59fe010); /* ldr lr, [pc, #16] */
1211 write32le(p+8, 0xe08fe00e); /* add lr, pc, lr */
1212 write32le(p+12, 0xe5bef008); /* ldr pc, [lr, #8]! */
1213 }
1214
1215 symattr->plt_offset = plt->data_offset;
1216 if (symattr->plt_thumb_stub) {
1217 p = section_ptr_add(plt, 20);
1218 write32le(p, 0x4778); /* bx pc */
1219 write32le(p+2, 0x46c0); /* nop */
1220 p += 4;
1221 } else
1222 p = section_ptr_add(plt, 16);
1223 write32le(p, 0xe59fc004); /* ldr ip, [pc, #4] ; GOT entry offset */
1224 write32le(p+4, 0xe08fc00c); /* add ip, pc, ip ; addr of GOT entry */
1225 write32le(p+8, 0xe59cf000); /* ldr pc, [ip] ; jump to GOT entry */
1226 write32le(p+12, s1->got->data_offset); /* GOT entry off once patched */
1227
1228 /* the symbol is modified so that it will be relocated to
1229 the PLT */
1230 if (sym->st_shndx == SHN_UNDEF)
1231 offset = plt->data_offset - 16;
1232 }
1233 #elif defined(TCC_TARGET_ARM64)
1234 if (need_plt_entry) {
1235 Section *plt;
1236 uint8_t *p;
1237
1238 if (s1->output_type == TCC_OUTPUT_DLL)
1239 tcc_error("DLLs unimplemented!");
1240
1241 plt = s1->plt;
1242 if (plt->data_offset == 0)
1243 section_ptr_add(plt, 32);
1244 symattr->plt_offset = plt->data_offset;
1245 p = section_ptr_add(plt, 16);
1246 write32le(p, s1->got->data_offset);
1247 write32le(p + 4, (uint64_t)s1->got->data_offset >> 32);
1248
1249 if (sym->st_shndx == SHN_UNDEF)
1250 offset = plt->data_offset - 16;
1251 }
1252 #elif defined(TCC_TARGET_C67)
1253 if (s1->dynsym) {
1254 tcc_error("C67 got not implemented");
1255 }
1256 #else
1257 #error unsupported CPU
1258 #endif
1259 if (s1->dynsym) {
1260 /* XXX This might generate multiple syms for name. */
1261 index = put_elf_sym(s1->dynsym, offset,
1262 size, info, 0, sym->st_shndx, name);
1263 /* Create the relocation (it's against the GOT for PLT
1264 and GOT relocs). */
1265 put_elf_reloc(s1->dynsym, s1->got,
1266 s1->got->data_offset,
1267 reloc_type, index);
1268 } else {
1269 /* Without .dynsym (i.e. static link or memory output) we
1270 still need relocs against the generated got, so as to fill
1271 the entries with the symbol values (determined later). */
1272 put_elf_reloc(symtab_section, s1->got,
1273 s1->got->data_offset,
1274 reloc_type, sym_index);
1275 }
1276 /* And now create the GOT slot itself. */
1277 ptr = section_ptr_add(s1->got, PTR_SIZE);
1278 *ptr = 0;
1279 if (need_plt_entry)
1280 return symattr->plt_offset;
1281 else
1282 return symattr->got_offset;
1283 }
1284
1285 /* build GOT and PLT entries */
1286 ST_FUNC void build_got_entries(TCCState *s1)
1287 {
1288 Section *s;
1289 ElfW_Rel *rel;
1290 ElfW(Sym) *sym;
1291 int i, type, reloc_type, sym_index;
1292
1293 for(i = 1; i < s1->nb_sections; i++) {
1294 s = s1->sections[i];
1295 if (s->sh_type != SHT_RELX)
1296 continue;
1297 /* no need to handle got relocations */
1298 if (s->link != symtab_section)
1299 continue;
1300 for_each_elem(s, 0, rel, ElfW_Rel) {
1301 type = ELFW(R_TYPE)(rel->r_info);
1302 switch(type) {
1303 #if defined(TCC_TARGET_I386)
1304 case R_386_GOT32:
1305 case R_386_GOT32X:
1306 case R_386_GOTOFF:
1307 case R_386_GOTPC:
1308 case R_386_PLT32:
1309 if (!s1->got)
1310 build_got(s1);
1311 if (type == R_386_GOT32 || type == R_386_GOT32X ||
1312 type == R_386_PLT32) {
1313 sym_index = ELFW(R_SYM)(rel->r_info);
1314 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1315 /* look at the symbol got offset. If none, then add one */
1316 if (type == R_386_GOT32 || type == R_386_GOT32X)
1317 reloc_type = R_386_GLOB_DAT;
1318 else
1319 reloc_type = R_386_JMP_SLOT;
1320 put_got_entry(s1, reloc_type, sym->st_size, sym->st_info,
1321 sym_index);
1322 }
1323 break;
1324 #elif defined(TCC_TARGET_ARM)
1325 case R_ARM_PC24:
1326 case R_ARM_CALL:
1327 case R_ARM_JUMP24:
1328 case R_ARM_GOT32:
1329 case R_ARM_GOTOFF:
1330 case R_ARM_GOTPC:
1331 case R_ARM_PLT32:
1332 if (!s1->got)
1333 build_got(s1);
1334 sym_index = ELFW(R_SYM)(rel->r_info);
1335 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1336 if (type != R_ARM_GOTOFF && type != R_ARM_GOTPC
1337 && sym->st_shndx == SHN_UNDEF) {
1338 unsigned long ofs;
1339 /* look at the symbol got offset. If none, then add one */
1340 if (type == R_ARM_GOT32)
1341 reloc_type = R_ARM_GLOB_DAT;
1342 else
1343 reloc_type = R_ARM_JUMP_SLOT;
1344 ofs = put_got_entry(s1, reloc_type, sym->st_size,
1345 sym->st_info, sym_index);
1346 #ifdef DEBUG_RELOC
1347 printf ("maybegot: %s, %d, %d --> ofs=0x%x\n",
1348 (char *) symtab_section->link->data + sym->st_name,
1349 type, sym->st_shndx, ofs);
1350 #endif
1351 if (type != R_ARM_GOT32) {
1352 addr_t *ptr = (addr_t*)(s1->sections[s->sh_info]->data
1353 + rel->r_offset);
1354 /* x must be signed! */
1355 int x = *ptr & 0xffffff;
1356 x = (x << 8) >> 8;
1357 x <<= 2;
1358 x += ofs;
1359 x >>= 2;
1360 #ifdef DEBUG_RELOC
1361 printf ("insn=0x%x --> 0x%x (x==0x%x)\n", *ptr,
1362 (*ptr & 0xff000000) | x, x);
1363 #endif
1364 *ptr = (*ptr & 0xff000000) | x;
1365 }
1366 }
1367 break;
1368 case R_ARM_THM_JUMP24:
1369 sym_index = ELFW(R_SYM)(rel->r_info);
1370 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1371 /* We are relocating a jump from thumb code to arm code */
1372 if (sym->st_shndx != SHN_UNDEF && !(sym->st_value & 1)) {
1373 int index;
1374 uint8_t *p;
1375 char *name, buf[1024];
1376 Section *text_section;
1377
1378 name = (char *) symtab_section->link->data + sym->st_name;
1379 text_section = s1->sections[sym->st_shndx];
1380 /* Modify reloc to target a thumb stub to switch to ARM */
1381 snprintf(buf, sizeof(buf), "%s_from_thumb", name);
1382 index = put_elf_sym(symtab_section,
1383 text_section->data_offset + 1,
1384 sym->st_size, sym->st_info, 0,
1385 sym->st_shndx, buf);
1386 rel->r_info = ELFW(R_INFO)(index, type);
1387 /* Create a thumb stub fonction to switch to ARM mode */
1388 put_elf_reloc(symtab_section, text_section,
1389 text_section->data_offset + 4, R_ARM_JUMP24,
1390 sym_index);
1391 p = section_ptr_add(text_section, 8);
1392 write32le(p, 0x4778); /* bx pc */
1393 write32le(p+2, 0x46c0); /* nop */
1394 write32le(p+4, 0xeafffffe); /* b $sym */
1395 }
1396 #elif defined(TCC_TARGET_ARM64)
1397 //xx Other cases may be required here:
1398 case R_AARCH64_ADR_GOT_PAGE:
1399 case R_AARCH64_LD64_GOT_LO12_NC:
1400 if (!s1->got)
1401 build_got(s1);
1402 sym_index = ELFW(R_SYM)(rel->r_info);
1403 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1404 reloc_type = R_AARCH64_GLOB_DAT;
1405 put_got_entry(s1, reloc_type, sym->st_size, sym->st_info,
1406 sym_index);
1407 break;
1408
1409 case R_AARCH64_JUMP26:
1410 case R_AARCH64_CALL26:
1411 if (!s1->got)
1412 build_got(s1);
1413 sym_index = ELFW(R_SYM)(rel->r_info);
1414 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1415 if (sym->st_shndx == SHN_UNDEF) {
1416 unsigned long ofs;
1417 reloc_type = R_AARCH64_JUMP_SLOT;
1418 ofs = put_got_entry(s1, reloc_type, sym->st_size,
1419 sym->st_info, sym_index);
1420 /* We store the place of the generated PLT slot
1421 in our addend. */
1422 rel->r_addend += ofs;
1423 }
1424 break;
1425 #elif defined(TCC_TARGET_C67)
1426 case R_C60_GOT32:
1427 case R_C60_GOTOFF:
1428 case R_C60_GOTPC:
1429 case R_C60_PLT32:
1430 if (!s1->got)
1431 build_got(s1);
1432 if (type == R_C60_GOT32 || type == R_C60_PLT32) {
1433 sym_index = ELFW(R_SYM)(rel->r_info);
1434 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1435 /* look at the symbol got offset. If none, then add one */
1436 if (type == R_C60_GOT32)
1437 reloc_type = R_C60_GLOB_DAT;
1438 else
1439 reloc_type = R_C60_JMP_SLOT;
1440 put_got_entry(s1, reloc_type, sym->st_size, sym->st_info,
1441 sym_index);
1442 }
1443 break;
1444 #elif defined(TCC_TARGET_X86_64)
1445 case R_X86_64_GOT32:
1446 case R_X86_64_GOTTPOFF:
1447 case R_X86_64_GOTPCREL:
1448 case R_X86_64_GOTPCRELX:
1449 case R_X86_64_REX_GOTPCRELX:
1450 case R_X86_64_PLT32:
1451 sym_index = ELFW(R_SYM)(rel->r_info);
1452 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1453 if (type == R_X86_64_PLT32 &&
1454 ELFW(ST_VISIBILITY)(sym->st_other) != STV_DEFAULT)
1455 {
1456 rel->r_info = ELFW(R_INFO)(sym_index, R_X86_64_PC32);
1457 break;
1458 }
1459
1460 if (!s1->got) {
1461 build_got(s1);
1462 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1463 }
1464 if (type == R_X86_64_GOT32 || type == R_X86_64_GOTPCREL ||
1465 type == R_X86_64_GOTPCRELX ||
1466 type == R_X86_64_REX_GOTPCRELX ||
1467 type == R_X86_64_PLT32) {
1468 unsigned long ofs;
1469 /* look at the symbol got offset. If none, then add one */
1470 if (type == R_X86_64_PLT32)
1471 reloc_type = R_X86_64_JUMP_SLOT;
1472 else
1473 reloc_type = R_X86_64_GLOB_DAT;
1474 ofs = put_got_entry(s1, reloc_type, sym->st_size,
1475 sym->st_info, sym_index);
1476 if (type == R_X86_64_PLT32)
1477 /* We store the place of the generated PLT slot
1478 in our addend. */
1479 rel->r_addend += ofs;
1480 }
1481 break;
1482 #else
1483 #error unsupported CPU
1484 #endif
1485 default:
1486 break;
1487 }
1488 }
1489 }
1490 }
1491
1492 ST_FUNC Section *new_symtab(TCCState *s1,
1493 const char *symtab_name, int sh_type, int sh_flags,
1494 const char *strtab_name,
1495 const char *hash_name, int hash_sh_flags)
1496 {
1497 Section *symtab, *strtab, *hash;
1498 int *ptr, nb_buckets;
1499
1500 symtab = new_section(s1, symtab_name, sh_type, sh_flags);
1501 symtab->sh_entsize = sizeof(ElfW(Sym));
1502 strtab = new_section(s1, strtab_name, SHT_STRTAB, sh_flags);
1503 put_elf_str(strtab, "");
1504 symtab->link = strtab;
1505 put_elf_sym(symtab, 0, 0, 0, 0, 0, NULL);
1506
1507 nb_buckets = 1;
1508
1509 hash = new_section(s1, hash_name, SHT_HASH, hash_sh_flags);
1510 hash->sh_entsize = sizeof(int);
1511 symtab->hash = hash;
1512 hash->link = symtab;
1513
1514 ptr = section_ptr_add(hash, (2 + nb_buckets + 1) * sizeof(int));
1515 ptr[0] = nb_buckets;
1516 ptr[1] = 1;
1517 memset(ptr + 2, 0, (nb_buckets + 1) * sizeof(int));
1518 return symtab;
1519 }
1520
1521 /* put dynamic tag */
1522 static void put_dt(Section *dynamic, int dt, addr_t val)
1523 {
1524 ElfW(Dyn) *dyn;
1525 dyn = section_ptr_add(dynamic, sizeof(ElfW(Dyn)));
1526 dyn->d_tag = dt;
1527 dyn->d_un.d_val = val;
1528 }
1529
1530 static void add_init_array_defines(TCCState *s1, const char *section_name)
1531 {
1532 Section *s;
1533 long end_offset;
1534 char sym_start[1024];
1535 char sym_end[1024];
1536
1537 snprintf(sym_start, sizeof(sym_start), "__%s_start", section_name + 1);
1538 snprintf(sym_end, sizeof(sym_end), "__%s_end", section_name + 1);
1539
1540 s = find_section(s1, section_name);
1541 if (!s) {
1542 end_offset = 0;
1543 s = data_section;
1544 } else {
1545 end_offset = s->data_offset;
1546 }
1547
1548 add_elf_sym(symtab_section,
1549 0, 0,
1550 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1551 s->sh_num, sym_start);
1552 add_elf_sym(symtab_section,
1553 end_offset, 0,
1554 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1555 s->sh_num, sym_end);
1556 }
1557
1558 static int tcc_add_support(TCCState *s1, const char *filename)
1559 {
1560 char buf[1024];
1561 snprintf(buf, sizeof(buf), "%s/"TCC_ARCH_DIR"%s", s1->tcc_lib_path, filename);
1562 return tcc_add_file(s1, buf);
1563 }
1564
1565 ST_FUNC void tcc_add_bcheck(TCCState *s1)
1566 {
1567 #ifdef CONFIG_TCC_BCHECK
1568 addr_t *ptr;
1569 int sym_index;
1570
1571 if (0 == s1->do_bounds_check)
1572 return;
1573 /* XXX: add an object file to do that */
1574 ptr = section_ptr_add(bounds_section, sizeof(*ptr));
1575 *ptr = 0;
1576 add_elf_sym(symtab_section, 0, 0,
1577 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1578 bounds_section->sh_num, "__bounds_start");
1579 /* pull bcheck.o from libtcc1.a */
1580 sym_index = add_elf_sym(symtab_section, 0, 0,
1581 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1582 SHN_UNDEF, "__bound_init");
1583 if (s1->output_type != TCC_OUTPUT_MEMORY) {
1584 /* add 'call __bound_init()' in .init section */
1585 Section *init_section = find_section(s1, ".init");
1586 unsigned char *pinit = section_ptr_add(init_section, 5);
1587 pinit[0] = 0xe8;
1588 write32le(pinit + 1, -4);
1589 put_elf_reloc(symtab_section, init_section,
1590 init_section->data_offset - 4, R_386_PC32, sym_index);
1591 /* R_386_PC32 = R_X86_64_PC32 = 2 */
1592 }
1593 #endif
1594 }
1595
1596 /* add tcc runtime libraries */
1597 ST_FUNC void tcc_add_runtime(TCCState *s1)
1598 {
1599 tcc_add_bcheck(s1);
1600 tcc_add_pragma_libs(s1);
1601 /* add libc */
1602 if (!s1->nostdlib) {
1603 tcc_add_library(s1, "c");
1604 #ifdef CONFIG_USE_LIBGCC
1605 if (!s1->static_link) {
1606 tcc_add_file(s1, TCC_LIBGCC);
1607 }
1608 #endif
1609 tcc_add_support(s1, "libtcc1.a");
1610 /* add crt end if not memory output */
1611 if (s1->output_type != TCC_OUTPUT_MEMORY)
1612 tcc_add_crt(s1, "crtn.o");
1613 }
1614 }
1615
1616 /* add various standard linker symbols (must be done after the
1617 sections are filled (for example after allocating common
1618 symbols)) */
1619 ST_FUNC void tcc_add_linker_symbols(TCCState *s1)
1620 {
1621 char buf[1024];
1622 int i;
1623 Section *s;
1624
1625 add_elf_sym(symtab_section,
1626 text_section->data_offset, 0,
1627 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1628 text_section->sh_num, "_etext");
1629 add_elf_sym(symtab_section,
1630 data_section->data_offset, 0,
1631 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1632 data_section->sh_num, "_edata");
1633 add_elf_sym(symtab_section,
1634 bss_section->data_offset, 0,
1635 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1636 bss_section->sh_num, "_end");
1637 /* horrible new standard ldscript defines */
1638 add_init_array_defines(s1, ".preinit_array");
1639 add_init_array_defines(s1, ".init_array");
1640 add_init_array_defines(s1, ".fini_array");
1641
1642 /* add start and stop symbols for sections whose name can be
1643 expressed in C */
1644 for(i = 1; i < s1->nb_sections; i++) {
1645 s = s1->sections[i];
1646 if (s->sh_type == SHT_PROGBITS &&
1647 (s->sh_flags & SHF_ALLOC)) {
1648 const char *p;
1649 int ch;
1650
1651 /* check if section name can be expressed in C */
1652 p = s->name;
1653 for(;;) {
1654 ch = *p;
1655 if (!ch)
1656 break;
1657 if (!isid(ch) && !isnum(ch))
1658 goto next_sec;
1659 p++;
1660 }
1661 snprintf(buf, sizeof(buf), "__start_%s", s->name);
1662 add_elf_sym(symtab_section,
1663 0, 0,
1664 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1665 s->sh_num, buf);
1666 snprintf(buf, sizeof(buf), "__stop_%s", s->name);
1667 add_elf_sym(symtab_section,
1668 s->data_offset, 0,
1669 ELFW(ST_INFO)(STB_GLOBAL, STT_NOTYPE), 0,
1670 s->sh_num, buf);
1671 }
1672 next_sec: ;
1673 }
1674 }
1675
1676 static void tcc_output_binary(TCCState *s1, FILE *f,
1677 const int *sec_order)
1678 {
1679 Section *s;
1680 int i, offset, size;
1681
1682 offset = 0;
1683 for(i=1;i<s1->nb_sections;i++) {
1684 s = s1->sections[sec_order[i]];
1685 if (s->sh_type != SHT_NOBITS &&
1686 (s->sh_flags & SHF_ALLOC)) {
1687 while (offset < s->sh_offset) {
1688 fputc(0, f);
1689 offset++;
1690 }
1691 size = s->sh_size;
1692 fwrite(s->data, 1, size, f);
1693 offset += size;
1694 }
1695 }
1696 }
1697
1698 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
1699 #define HAVE_PHDR 1
1700 #define EXTRA_RELITEMS 14
1701
1702 /* move the relocation value from .dynsym to .got */
1703 static void patch_dynsym_undef(TCCState *s1, Section *s)
1704 {
1705 uint32_t *gotd = (void *)s1->got->data;
1706 ElfW(Sym) *sym;
1707
1708 gotd += 3; /* dummy entries in .got */
1709 /* relocate symbols in .dynsym */
1710 for_each_elem(s, 1, sym, ElfW(Sym)) {
1711 if (sym->st_shndx == SHN_UNDEF) {
1712 *gotd++ = sym->st_value + 6; /* XXX 6 is magic ? */
1713 sym->st_value = 0;
1714 }
1715 }
1716 }
1717 #else
1718 #define HAVE_PHDR 1
1719 #define EXTRA_RELITEMS 9
1720
1721 /* zero plt offsets of weak symbols in .dynsym */
1722 static void patch_dynsym_undef(TCCState *s1, Section *s)
1723 {
1724 ElfW(Sym) *sym;
1725
1726 for_each_elem(s, 1, sym, ElfW(Sym))
1727 if (sym->st_shndx == SHN_UNDEF && ELFW(ST_BIND)(sym->st_info) == STB_WEAK)
1728 sym->st_value = 0;
1729 }
1730 #endif
1731
1732 ST_FUNC void fill_got_entry(TCCState *s1, ElfW_Rel *rel)
1733 {
1734 int sym_index = ELFW(R_SYM) (rel->r_info);
1735 ElfW(Sym) *sym = &((ElfW(Sym) *) symtab_section->data)[sym_index];
1736 unsigned long offset;
1737
1738 if (sym_index >= s1->nb_sym_attrs)
1739 return;
1740 offset = s1->sym_attrs[sym_index].got_offset;
1741 section_reserve(s1->got, offset + PTR_SIZE);
1742 #ifdef TCC_TARGET_X86_64
1743 /* only works for x86-64 */
1744 write32le(s1->got->data + offset + 4, sym->st_value >> 32);
1745 #endif
1746 write32le(s1->got->data + offset, sym->st_value & 0xffffffff);
1747 }
1748
1749 /* Perform relocation to GOT or PLT entries */
1750 ST_FUNC void fill_got(TCCState *s1)
1751 {
1752 Section *s;
1753 ElfW_Rel *rel;
1754 int i;
1755
1756 for(i = 1; i < s1->nb_sections; i++) {
1757 s = s1->sections[i];
1758 if (s->sh_type != SHT_RELX)
1759 continue;
1760 /* no need to handle got relocations */
1761 if (s->link != symtab_section)
1762 continue;
1763 for_each_elem(s, 0, rel, ElfW_Rel) {
1764 switch (ELFW(R_TYPE) (rel->r_info)) {
1765 case R_X86_64_GOT32:
1766 case R_X86_64_GOTPCREL:
1767 case R_X86_64_GOTPCRELX:
1768 case R_X86_64_REX_GOTPCRELX:
1769 case R_X86_64_PLT32:
1770 fill_got_entry(s1, rel);
1771 break;
1772 }
1773 }
1774 }
1775 }
1776
1777 /* Bind symbols of executable: resolve undefined symbols from exported symbols
1778 in shared libraries and export non local defined symbols to shared libraries
1779 if -rdynamic switch was given on command line */
1780 static void bind_exe_dynsyms(TCCState *s1)
1781 {
1782 const char *name;
1783 int sym_index, index;
1784 ElfW(Sym) *sym, *esym;
1785 int type;
1786
1787 /* Resolve undefined symbols from dynamic symbols. When there is a match:
1788 - if STT_FUNC or STT_GNU_IFUNC symbol -> add it in PLT
1789 - if STT_OBJECT symbol -> add it in .bss section with suitable reloc */
1790 for_each_elem(symtab_section, 1, sym, ElfW(Sym)) {
1791 if (sym->st_shndx == SHN_UNDEF) {
1792 name = (char *) symtab_section->link->data + sym->st_name;
1793 sym_index = find_elf_sym(s1->dynsymtab_section, name);
1794 if (sym_index) {
1795 esym = &((ElfW(Sym) *)s1->dynsymtab_section->data)[sym_index];
1796 type = ELFW(ST_TYPE)(esym->st_info);
1797 if ((type == STT_FUNC) || (type == STT_GNU_IFUNC)) {
1798 /* Indirect functions shall have STT_FUNC type in executable
1799 * dynsym section. Indeed, a dlsym call following a lazy
1800 * resolution would pick the symbol value from the
1801 * executable dynsym entry which would contain the address
1802 * of the function wanted by the caller of dlsym instead of
1803 * the address of the function that would return that
1804 * address */
1805 put_got_entry(s1, R_JMP_SLOT, esym->st_size,
1806 ELFW(ST_INFO)(STB_GLOBAL,STT_FUNC),
1807 sym - (ElfW(Sym) *)symtab_section->data);
1808 } else if (type == STT_OBJECT) {
1809 unsigned long offset;
1810 ElfW(Sym) *dynsym;
1811 offset = bss_section->data_offset;
1812 /* XXX: which alignment ? */
1813 offset = (offset + 16 - 1) & -16;
1814 index = put_elf_sym(s1->dynsym, offset, esym->st_size,
1815 esym->st_info, 0, bss_section->sh_num,
1816 name);
1817 /* Ensure R_COPY works for weak symbol aliases */
1818 if (ELFW(ST_BIND)(esym->st_info) == STB_WEAK) {
1819 for_each_elem(s1->dynsymtab_section, 1, dynsym, ElfW(Sym)) {
1820 if ((dynsym->st_value == esym->st_value)
1821 && (ELFW(ST_BIND)(dynsym->st_info) == STB_GLOBAL)) {
1822 char *dynname = (char *) s1->dynsymtab_section->link->data
1823 + dynsym->st_name;
1824 put_elf_sym(s1->dynsym, offset, dynsym->st_size,
1825 dynsym->st_info, 0,
1826 bss_section->sh_num, dynname);
1827 break;
1828 }
1829 }
1830 }
1831 put_elf_reloc(s1->dynsym, bss_section,
1832 offset, R_COPY, index);
1833 offset += esym->st_size;
1834 bss_section->data_offset = offset;
1835 }
1836 } else {
1837 /* STB_WEAK undefined symbols are accepted */
1838 /* XXX: _fp_hw seems to be part of the ABI, so we ignore it */
1839 if (ELFW(ST_BIND)(sym->st_info) == STB_WEAK ||
1840 !strcmp(name, "_fp_hw")) {
1841 } else {
1842 tcc_error_noabort("undefined symbol '%s'", name);
1843 }
1844 }
1845 } else if (s1->rdynamic && ELFW(ST_BIND)(sym->st_info) != STB_LOCAL) {
1846 /* if -rdynamic option, then export all non local symbols */
1847 name = (char *) symtab_section->link->data + sym->st_name;
1848 put_elf_sym(s1->dynsym, sym->st_value, sym->st_size, sym->st_info,
1849 0, sym->st_shndx, name);
1850 }
1851 }
1852 }
1853
1854 /* Bind symbols of libraries: export non local symbols of executable that
1855 resolve undefined symbols of shared libraries */
1856 static void bind_libs_dynsyms(TCCState *s1)
1857 {
1858 const char *name;
1859 int sym_index;
1860 ElfW(Sym) *sym, *esym;
1861
1862 /* now look at unresolved dynamic symbols and export
1863 corresponding symbol */
1864 for_each_elem(s1->dynsymtab_section, 1, esym, ElfW(Sym)) {
1865 name = (char *) s1->dynsymtab_section->link->data + esym->st_name;
1866 sym_index = find_elf_sym(symtab_section, name);
1867 if (sym_index) {
1868 /* XXX: avoid adding a symbol if already present because of
1869 -rdynamic ? */
1870 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
1871 if (sym->st_shndx != SHN_UNDEF)
1872 put_elf_sym(s1->dynsym, sym->st_value, sym->st_size,
1873 sym->st_info, 0, sym->st_shndx, name);
1874 } else if (esym->st_shndx == SHN_UNDEF) {
1875 /* weak symbols can stay undefined */
1876 if (ELFW(ST_BIND)(esym->st_info) != STB_WEAK)
1877 tcc_warning("undefined dynamic symbol '%s'", name);
1878 }
1879 }
1880 }
1881
1882 /* Export all non local symbols (for shared libraries) */
1883 static void export_global_syms(TCCState *s1)
1884 {
1885 int nb_syms, dynindex, index;
1886 const char *name;
1887 ElfW(Sym) *sym;
1888
1889 nb_syms = symtab_section->data_offset / sizeof(ElfW(Sym));
1890 s1->symtab_to_dynsym = tcc_mallocz(sizeof(int) * nb_syms);
1891 for_each_elem(symtab_section, 1, sym, ElfW(Sym)) {
1892 if (ELFW(ST_BIND)(sym->st_info) != STB_LOCAL) {
1893 name = (char *) symtab_section->link->data + sym->st_name;
1894 dynindex = put_elf_sym(s1->dynsym, sym->st_value, sym->st_size,
1895 sym->st_info, 0, sym->st_shndx, name);
1896 index = sym - (ElfW(Sym) *) symtab_section->data;
1897 s1->symtab_to_dynsym[index] = dynindex;
1898 }
1899 }
1900 }
1901
1902 /* relocate the PLT: compute addresses and offsets in the PLT now that final
1903 address for PLT and GOT are known (see fill_program_header) */
1904 ST_FUNC void relocate_plt(TCCState *s1)
1905 {
1906 uint8_t *p, *p_end;
1907
1908 if (!s1->plt)
1909 return;
1910
1911 p = s1->plt->data;
1912 p_end = p + s1->plt->data_offset;
1913 if (p < p_end) {
1914 #if defined(TCC_TARGET_I386)
1915 write32le(p + 2, read32le(p + 2) + s1->got->sh_addr);
1916 write32le(p + 8, read32le(p + 8) + s1->got->sh_addr);
1917 p += 16;
1918 while (p < p_end) {
1919 write32le(p + 2, read32le(p + 2) + s1->got->sh_addr);
1920 p += 16;
1921 }
1922 #elif defined(TCC_TARGET_X86_64)
1923 int x = s1->got->sh_addr - s1->plt->sh_addr - 6;
1924 write32le(p + 2, read32le(p + 2) + x);
1925 write32le(p + 8, read32le(p + 8) + x - 6);
1926 p += 16;
1927 while (p < p_end) {
1928 write32le(p + 2, read32le(p + 2) + x + s1->plt->data - p);
1929 p += 16;
1930 }
1931 #elif defined(TCC_TARGET_ARM)
1932 int x;
1933 x=s1->got->sh_addr - s1->plt->sh_addr - 12;
1934 p += 16;
1935 while (p < p_end) {
1936 if (read32le(p) == 0x46c04778) /* PLT Thumb stub present */
1937 p += 4;
1938 write32le(p + 12, x + read32le(p + 12) + s1->plt->data - p);
1939 p += 16;
1940 }
1941 #elif defined(TCC_TARGET_ARM64)
1942 uint64_t plt = s1->plt->sh_addr;
1943 uint64_t got = s1->got->sh_addr;
1944 uint64_t off = (got >> 12) - (plt >> 12);
1945 if ((off + ((uint32_t)1 << 20)) >> 21)
1946 tcc_error("Failed relocating PLT (off=0x%lx, got=0x%lx, plt=0x%lx)", off, got, plt);
1947 write32le(p, 0xa9bf7bf0); // stp x16,x30,[sp,#-16]!
1948 write32le(p + 4, (0x90000010 | // adrp x16,...
1949 (off & 0x1ffffc) << 3 | (off & 3) << 29));
1950 write32le(p + 8, (0xf9400211 | // ldr x17,[x16,#...]
1951 (got & 0xff8) << 7));
1952 write32le(p + 12, (0x91000210 | // add x16,x16,#...
1953 (got & 0xfff) << 10));
1954 write32le(p + 16, 0xd61f0220); // br x17
1955 write32le(p + 20, 0xd503201f); // nop
1956 write32le(p + 24, 0xd503201f); // nop
1957 write32le(p + 28, 0xd503201f); // nop
1958 p += 32;
1959 while (p < p_end) {
1960 uint64_t pc = plt + (p - s1->plt->data);
1961 uint64_t addr = got + read64le(p);
1962 uint64_t off = (addr >> 12) - (pc >> 12);
1963 if ((off + ((uint32_t)1 << 20)) >> 21)
1964 tcc_error("Failed relocating PLT (off=0x%lx, addr=0x%lx, pc=0x%lx)", off, addr, pc);
1965 write32le(p, (0x90000010 | // adrp x16,...
1966 (off & 0x1ffffc) << 3 | (off & 3) << 29));
1967 write32le(p + 4, (0xf9400211 | // ldr x17,[x16,#...]
1968 (addr & 0xff8) << 7));
1969 write32le(p + 8, (0x91000210 | // add x16,x16,#...
1970 (addr & 0xfff) << 10));
1971 write32le(p + 12, 0xd61f0220); // br x17
1972 p += 16;
1973 }
1974 #elif defined(TCC_TARGET_C67)
1975 /* XXX: TODO */
1976 #else
1977 #error unsupported CPU
1978 #endif
1979 }
1980 }
1981
1982 /* Allocate strings for section names and decide if an unallocated section
1983 should be output.
1984
1985 NOTE: the strsec section comes last, so its size is also correct ! */
1986 static void alloc_sec_names(TCCState *s1, int file_type, Section *strsec)
1987 {
1988 int i;
1989 Section *s;
1990
1991 /* Allocate strings for section names */
1992 for(i = 1; i < s1->nb_sections; i++) {
1993 s = s1->sections[i];
1994 s->sh_name = put_elf_str(strsec, s->name);
1995 /* when generating a DLL, we include relocations but we may
1996 patch them */
1997 if (file_type == TCC_OUTPUT_DLL &&
1998 s->sh_type == SHT_RELX &&
1999 !(s->sh_flags & SHF_ALLOC)) {
2000 /* gr: avoid bogus relocs for empty (debug) sections */
2001 if (s1->sections[s->sh_info]->sh_flags & SHF_ALLOC)
2002 prepare_dynamic_rel(s1, s);
2003 else if (s1->do_debug)
2004 s->sh_size = s->data_offset;
2005 } else if (s1->do_debug ||
2006 file_type == TCC_OUTPUT_OBJ ||
2007 (s->sh_flags & SHF_ALLOC) ||
2008 i == (s1->nb_sections - 1)) {
2009 /* we output all sections if debug or object file */
2010 s->sh_size = s->data_offset;
2011 }
2012 }
2013 }
2014
2015 /* Info to be copied in dynamic section */
2016 struct dyn_inf {
2017 Section *dynamic;
2018 Section *dynstr;
2019 unsigned long dyn_rel_off;
2020 addr_t rel_addr;
2021 addr_t rel_size;
2022 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
2023 addr_t bss_addr;
2024 addr_t bss_size;
2025 #endif
2026 };
2027
2028 /* Assign sections to segments and decide how are sections laid out when loaded
2029 in memory. This function also fills corresponding program headers. */
2030 static int layout_sections(TCCState *s1, ElfW(Phdr) *phdr, int phnum,
2031 Section *interp, Section* strsec,
2032 struct dyn_inf *dyninf, int *sec_order)
2033 {
2034 int i, j, k, file_type, sh_order_index, file_offset;
2035 unsigned long s_align;
2036 long long tmp;
2037 addr_t addr;
2038 ElfW(Phdr) *ph;
2039 Section *s;
2040
2041 file_type = s1->output_type;
2042 sh_order_index = 1;
2043 file_offset = 0;
2044 if (s1->output_format == TCC_OUTPUT_FORMAT_ELF)
2045 file_offset = sizeof(ElfW(Ehdr)) + phnum * sizeof(ElfW(Phdr));
2046 s_align = ELF_PAGE_SIZE;
2047 if (s1->section_align)
2048 s_align = s1->section_align;
2049
2050 if (phnum > 0) {
2051 if (s1->has_text_addr) {
2052 int a_offset, p_offset;
2053 addr = s1->text_addr;
2054 /* we ensure that (addr % ELF_PAGE_SIZE) == file_offset %
2055 ELF_PAGE_SIZE */
2056 a_offset = (int) (addr & (s_align - 1));
2057 p_offset = file_offset & (s_align - 1);
2058 if (a_offset < p_offset)
2059 a_offset += s_align;
2060 file_offset += (a_offset - p_offset);
2061 } else {
2062 if (file_type == TCC_OUTPUT_DLL)
2063 addr = 0;
2064 else
2065 addr = ELF_START_ADDR;
2066 /* compute address after headers */
2067 addr += (file_offset & (s_align - 1));
2068 }
2069
2070 ph = &phdr[0];
2071 /* Leave one program headers for the program interpreter and one for
2072 the program header table itself if needed. These are done later as
2073 they require section layout to be done first. */
2074 if (interp)
2075 ph += 1 + HAVE_PHDR;
2076
2077 /* dynamic relocation table information, for .dynamic section */
2078 dyninf->rel_addr = dyninf->rel_size = 0;
2079 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
2080 dyninf->bss_addr = dyninf->bss_size = 0;
2081 #endif
2082
2083 for(j = 0; j < 2; j++) {
2084 ph->p_type = PT_LOAD;
2085 if (j == 0)
2086 ph->p_flags = PF_R | PF_X;
2087 else
2088 ph->p_flags = PF_R | PF_W;
2089 ph->p_align = s_align;
2090
2091 /* Decide the layout of sections loaded in memory. This must
2092 be done before program headers are filled since they contain
2093 info about the layout. We do the following ordering: interp,
2094 symbol tables, relocations, progbits, nobits */
2095 /* XXX: do faster and simpler sorting */
2096 for(k = 0; k < 5; k++) {
2097 for(i = 1; i < s1->nb_sections; i++) {
2098 s = s1->sections[i];
2099 /* compute if section should be included */
2100 if (j == 0) {
2101 if ((s->sh_flags & (SHF_ALLOC | SHF_WRITE)) !=
2102 SHF_ALLOC)
2103 continue;
2104 } else {
2105 if ((s->sh_flags & (SHF_ALLOC | SHF_WRITE)) !=
2106 (SHF_ALLOC | SHF_WRITE))
2107 continue;
2108 }
2109 if (s == interp) {
2110 if (k != 0)
2111 continue;
2112 } else if (s->sh_type == SHT_DYNSYM ||
2113 s->sh_type == SHT_STRTAB ||
2114 s->sh_type == SHT_HASH) {
2115 if (k != 1)
2116 continue;
2117 } else if (s->sh_type == SHT_RELX) {
2118 if (k != 2)
2119 continue;
2120 } else if (s->sh_type == SHT_NOBITS) {
2121 if (k != 4)
2122 continue;
2123 } else {
2124 if (k != 3)
2125 continue;
2126 }
2127 sec_order[sh_order_index++] = i;
2128
2129 /* section matches: we align it and add its size */
2130 tmp = addr;
2131 addr = (addr + s->sh_addralign - 1) &
2132 ~(s->sh_addralign - 1);
2133 file_offset += (int) ( addr - tmp );
2134 s->sh_offset = file_offset;
2135 s->sh_addr = addr;
2136
2137 /* update program header infos */
2138 if (ph->p_offset == 0) {
2139 ph->p_offset = file_offset;
2140 ph->p_vaddr = addr;
2141 ph->p_paddr = ph->p_vaddr;
2142 }
2143 /* update dynamic relocation infos */
2144 if (s->sh_type == SHT_RELX) {
2145 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
2146 if (!strcmp(strsec->data + s->sh_name, ".rel.got")) {
2147 dyninf->rel_addr = addr;
2148 dyninf->rel_size += s->sh_size; /* XXX only first rel. */
2149 }
2150 if (!strcmp(strsec->data + s->sh_name, ".rel.bss")) {
2151 dyninf->bss_addr = addr;
2152 dyninf->bss_size = s->sh_size; /* XXX only first rel. */
2153 }
2154 #else
2155 if (dyninf->rel_size == 0)
2156 dyninf->rel_addr = addr;
2157 dyninf->rel_size += s->sh_size;
2158 #endif
2159 }
2160 addr += s->sh_size;
2161 if (s->sh_type != SHT_NOBITS)
2162 file_offset += s->sh_size;
2163 }
2164 }
2165 if (j == 0) {
2166 /* Make the first PT_LOAD segment include the program
2167 headers itself (and the ELF header as well), it'll
2168 come out with same memory use but will make various
2169 tools like binutils strip work better. */
2170 ph->p_offset &= ~(ph->p_align - 1);
2171 ph->p_vaddr &= ~(ph->p_align - 1);
2172 ph->p_paddr &= ~(ph->p_align - 1);
2173 }
2174 ph->p_filesz = file_offset - ph->p_offset;
2175 ph->p_memsz = addr - ph->p_vaddr;
2176 ph++;
2177 if (j == 0) {
2178 if (s1->output_format == TCC_OUTPUT_FORMAT_ELF) {
2179 /* if in the middle of a page, we duplicate the page in
2180 memory so that one copy is RX and the other is RW */
2181 if ((addr & (s_align - 1)) != 0)
2182 addr += s_align;
2183 } else {
2184 addr = (addr + s_align - 1) & ~(s_align - 1);
2185 file_offset = (file_offset + s_align - 1) & ~(s_align - 1);
2186 }
2187 }
2188 }
2189 }
2190
2191 /* all other sections come after */
2192 for(i = 1; i < s1->nb_sections; i++) {
2193 s = s1->sections[i];
2194 if (phnum > 0 && (s->sh_flags & SHF_ALLOC))
2195 continue;
2196 sec_order[sh_order_index++] = i;
2197
2198 file_offset = (file_offset + s->sh_addralign - 1) &
2199 ~(s->sh_addralign - 1);
2200 s->sh_offset = file_offset;
2201 if (s->sh_type != SHT_NOBITS)
2202 file_offset += s->sh_size;
2203 }
2204
2205 return file_offset;
2206 }
2207
2208 static void fill_unloadable_phdr(ElfW(Phdr) *phdr, int phnum, Section *interp,
2209 Section *dynamic)
2210 {
2211 ElfW(Phdr) *ph;
2212
2213 /* if interpreter, then add corresponding program header */
2214 if (interp) {
2215 ph = &phdr[0];
2216
2217 if (HAVE_PHDR)
2218 {
2219 int len = phnum * sizeof(ElfW(Phdr));
2220
2221 ph->p_type = PT_PHDR;
2222 ph->p_offset = sizeof(ElfW(Ehdr));
2223 ph->p_vaddr = interp->sh_addr - len;
2224 ph->p_paddr = ph->p_vaddr;
2225 ph->p_filesz = ph->p_memsz = len;
2226 ph->p_flags = PF_R | PF_X;
2227 ph->p_align = 4; /* interp->sh_addralign; */
2228 ph++;
2229 }
2230
2231 ph->p_type = PT_INTERP;
2232 ph->p_offset = interp->sh_offset;
2233 ph->p_vaddr = interp->sh_addr;
2234 ph->p_paddr = ph->p_vaddr;
2235 ph->p_filesz = interp->sh_size;
2236 ph->p_memsz = interp->sh_size;
2237 ph->p_flags = PF_R;
2238 ph->p_align = interp->sh_addralign;
2239 }
2240
2241 /* if dynamic section, then add corresponding program header */
2242 if (dynamic) {
2243 ph = &phdr[phnum - 1];
2244
2245 ph->p_type = PT_DYNAMIC;
2246 ph->p_offset = dynamic->sh_offset;
2247 ph->p_vaddr = dynamic->sh_addr;
2248 ph->p_paddr = ph->p_vaddr;
2249 ph->p_filesz = dynamic->sh_size;
2250 ph->p_memsz = dynamic->sh_size;
2251 ph->p_flags = PF_R | PF_W;
2252 ph->p_align = dynamic->sh_addralign;
2253 }
2254 }
2255
2256 /* Fill the dynamic section with tags describing the address and size of
2257 sections */
2258 static void fill_dynamic(TCCState *s1, struct dyn_inf *dyninf)
2259 {
2260 Section *dynamic;
2261
2262 dynamic = dyninf->dynamic;
2263
2264 /* put dynamic section entries */
2265 dynamic->data_offset = dyninf->dyn_rel_off;
2266 put_dt(dynamic, DT_HASH, s1->dynsym->hash->sh_addr);
2267 put_dt(dynamic, DT_STRTAB, dyninf->dynstr->sh_addr);
2268 put_dt(dynamic, DT_SYMTAB, s1->dynsym->sh_addr);
2269 put_dt(dynamic, DT_STRSZ, dyninf->dynstr->data_offset);
2270 put_dt(dynamic, DT_SYMENT, sizeof(ElfW(Sym)));
2271 #if defined(TCC_TARGET_ARM64) || defined(TCC_TARGET_X86_64)
2272 put_dt(dynamic, DT_RELA, dyninf->rel_addr);
2273 put_dt(dynamic, DT_RELASZ, dyninf->rel_size);
2274 put_dt(dynamic, DT_RELAENT, sizeof(ElfW_Rel));
2275 #else
2276 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
2277 put_dt(dynamic, DT_PLTGOT, s1->got->sh_addr);
2278 put_dt(dynamic, DT_PLTRELSZ, dyninf->rel_size);
2279 put_dt(dynamic, DT_JMPREL, dyninf->rel_addr);
2280 put_dt(dynamic, DT_PLTREL, DT_REL);
2281 put_dt(dynamic, DT_REL, dyninf->bss_addr);
2282 put_dt(dynamic, DT_RELSZ, dyninf->bss_size);
2283 #else
2284 put_dt(dynamic, DT_REL, dyninf->rel_addr);
2285 put_dt(dynamic, DT_RELSZ, dyninf->rel_size);
2286 put_dt(dynamic, DT_RELENT, sizeof(ElfW_Rel));
2287 #endif
2288 #endif
2289 if (s1->do_debug)
2290 put_dt(dynamic, DT_DEBUG, 0);
2291 put_dt(dynamic, DT_NULL, 0);
2292 }
2293
2294 /* Relocate remaining sections and symbols (that is those not related to
2295 dynamic linking) */
2296 static int final_sections_reloc(TCCState *s1)
2297 {
2298 int i;
2299 Section *s;
2300
2301 relocate_syms(s1, 0);
2302
2303 if (s1->nb_errors != 0)
2304 return -1;
2305
2306 /* relocate sections */
2307 /* XXX: ignore sections with allocated relocations ? */
2308 for(i = 1; i < s1->nb_sections; i++) {
2309 s = s1->sections[i];
2310 #ifdef TCC_TARGET_I386
2311 if (s->reloc && s != s1->got && (s->sh_flags & SHF_ALLOC)) //gr
2312 /* On X86 gdb 7.3 works in any case but gdb 6.6 will crash if SHF_ALLOC
2313 checking is removed */
2314 #else
2315 if (s->reloc && s != s1->got)
2316 /* On X86_64 gdb 7.3 will crash if SHF_ALLOC checking is present */
2317 #endif
2318 relocate_section(s1, s);
2319 }
2320
2321 /* relocate relocation entries if the relocation tables are
2322 allocated in the executable */
2323 for(i = 1; i < s1->nb_sections; i++) {
2324 s = s1->sections[i];
2325 if ((s->sh_flags & SHF_ALLOC) &&
2326 s->sh_type == SHT_RELX) {
2327 relocate_rel(s1, s);
2328 }
2329 }
2330 return 0;
2331 }
2332
2333 /* Create an ELF file on disk.
2334 This function handle ELF specific layout requirements */
2335 static void tcc_output_elf(TCCState *s1, FILE *f, int phnum, ElfW(Phdr) *phdr,
2336 int file_offset, int *sec_order)
2337 {
2338 int i, shnum, offset, size, file_type;
2339 Section *s;
2340 ElfW(Ehdr) ehdr;
2341 ElfW(Shdr) shdr, *sh;
2342
2343 file_type = s1->output_type;
2344 shnum = s1->nb_sections;
2345
2346 memset(&ehdr, 0, sizeof(ehdr));
2347
2348 if (phnum > 0) {
2349 ehdr.e_phentsize = sizeof(ElfW(Phdr));
2350 ehdr.e_phnum = phnum;
2351 ehdr.e_phoff = sizeof(ElfW(Ehdr));
2352 }
2353
2354 /* align to 4 */
2355 file_offset = (file_offset + 3) & -4;
2356
2357 /* fill header */
2358 ehdr.e_ident[0] = ELFMAG0;
2359 ehdr.e_ident[1] = ELFMAG1;
2360 ehdr.e_ident[2] = ELFMAG2;
2361 ehdr.e_ident[3] = ELFMAG3;
2362 ehdr.e_ident[4] = ELFCLASSW;
2363 ehdr.e_ident[5] = ELFDATA2LSB;
2364 ehdr.e_ident[6] = EV_CURRENT;
2365 #if defined(__FreeBSD__) || defined(__FreeBSD_kernel__)
2366 ehdr.e_ident[EI_OSABI] = ELFOSABI_FREEBSD;
2367 #endif
2368 #ifdef TCC_TARGET_ARM
2369 #ifdef TCC_ARM_EABI
2370 ehdr.e_ident[EI_OSABI] = 0;
2371 ehdr.e_flags = EF_ARM_EABI_VER4;
2372 if (file_type == TCC_OUTPUT_EXE || file_type == TCC_OUTPUT_DLL)
2373 ehdr.e_flags |= EF_ARM_HASENTRY;
2374 if (s1->float_abi == ARM_HARD_FLOAT)
2375 ehdr.e_flags |= EF_ARM_VFP_FLOAT;
2376 else
2377 ehdr.e_flags |= EF_ARM_SOFT_FLOAT;
2378 #else
2379 ehdr.e_ident[EI_OSABI] = ELFOSABI_ARM;
2380 #endif
2381 #endif
2382 switch(file_type) {
2383 default:
2384 case TCC_OUTPUT_EXE:
2385 ehdr.e_type = ET_EXEC;
2386 ehdr.e_entry = get_elf_sym_addr(s1, "_start", 1);
2387 break;
2388 case TCC_OUTPUT_DLL:
2389 ehdr.e_type = ET_DYN;
2390 ehdr.e_entry = text_section->sh_addr; /* XXX: is it correct ? */
2391 break;
2392 case TCC_OUTPUT_OBJ:
2393 ehdr.e_type = ET_REL;
2394 break;
2395 }
2396 ehdr.e_machine = EM_TCC_TARGET;
2397 ehdr.e_version = EV_CURRENT;
2398 ehdr.e_shoff = file_offset;
2399 ehdr.e_ehsize = sizeof(ElfW(Ehdr));
2400 ehdr.e_shentsize = sizeof(ElfW(Shdr));
2401 ehdr.e_shnum = shnum;
2402 ehdr.e_shstrndx = shnum - 1;
2403
2404 fwrite(&ehdr, 1, sizeof(ElfW(Ehdr)), f);
2405 fwrite(phdr, 1, phnum * sizeof(ElfW(Phdr)), f);
2406 offset = sizeof(ElfW(Ehdr)) + phnum * sizeof(ElfW(Phdr));
2407
2408 sort_syms(s1, symtab_section);
2409 for(i = 1; i < s1->nb_sections; i++) {
2410 s = s1->sections[sec_order[i]];
2411 if (s->sh_type != SHT_NOBITS) {
2412 if (s->sh_type == SHT_DYNSYM)
2413 patch_dynsym_undef(s1, s);
2414 while (offset < s->sh_offset) {
2415 fputc(0, f);
2416 offset++;
2417 }
2418 size = s->sh_size;
2419 if (size)
2420 fwrite(s->data, 1, size, f);
2421 offset += size;
2422 }
2423 }
2424
2425 /* output section headers */
2426 while (offset < ehdr.e_shoff) {
2427 fputc(0, f);
2428 offset++;
2429 }
2430
2431 for(i = 0; i < s1->nb_sections; i++) {
2432 sh = &shdr;
2433 memset(sh, 0, sizeof(ElfW(Shdr)));
2434 s = s1->sections[i];
2435 if (s) {
2436 sh->sh_name = s->sh_name;
2437 sh->sh_type = s->sh_type;
2438 sh->sh_flags = s->sh_flags;
2439 sh->sh_entsize = s->sh_entsize;
2440 sh->sh_info = s->sh_info;
2441 if (s->link)
2442 sh->sh_link = s->link->sh_num;
2443 sh->sh_addralign = s->sh_addralign;
2444 sh->sh_addr = s->sh_addr;
2445 sh->sh_offset = s->sh_offset;
2446 sh->sh_size = s->sh_size;
2447 }
2448 fwrite(sh, 1, sizeof(ElfW(Shdr)), f);
2449 }
2450 }
2451
2452 /* Write an elf, coff or "binary" file */
2453 static int tcc_write_elf_file(TCCState *s1, const char *filename, int phnum,
2454 ElfW(Phdr) *phdr, int file_offset, int *sec_order)
2455 {
2456 int fd, mode, file_type;
2457 FILE *f;
2458
2459 file_type = s1->output_type;
2460 if (file_type == TCC_OUTPUT_OBJ)
2461 mode = 0666;
2462 else
2463 mode = 0777;
2464 unlink(filename);
2465 fd = open(filename, O_WRONLY | O_CREAT | O_TRUNC | O_BINARY, mode);
2466 if (fd < 0) {
2467 tcc_error_noabort("could not write '%s'", filename);
2468 return -1;
2469 }
2470 f = fdopen(fd, "wb");
2471 if (s1->verbose)
2472 printf("<- %s\n", filename);
2473
2474 #ifdef TCC_TARGET_COFF
2475 if (s1->output_format == TCC_OUTPUT_FORMAT_COFF)
2476 tcc_output_coff(s1, f);
2477 else
2478 #endif
2479 if (s1->output_format == TCC_OUTPUT_FORMAT_ELF)
2480 tcc_output_elf(s1, f, phnum, phdr, file_offset, sec_order);
2481 else
2482 tcc_output_binary(s1, f, sec_order);
2483 fclose(f);
2484
2485 return 0;
2486 }
2487
2488 /* Output an elf, coff or binary file */
2489 /* XXX: suppress unneeded sections */
2490 static int elf_output_file(TCCState *s1, const char *filename)
2491 {
2492 int i, ret, phnum, shnum, file_type, file_offset, *sec_order;
2493 struct dyn_inf dyninf;
2494 ElfW(Phdr) *phdr;
2495 ElfW(Sym) *sym;
2496 Section *strsec, *interp, *dynamic, *dynstr;
2497
2498 file_type = s1->output_type;
2499 s1->nb_errors = 0;
2500
2501 /* if linking, also link in runtime libraries (libc, libgcc, etc.) */
2502 if (file_type != TCC_OUTPUT_OBJ) {
2503 tcc_add_runtime(s1);
2504 }
2505
2506 phdr = NULL;
2507 sec_order = NULL;
2508 interp = dynamic = dynstr = NULL; /* avoid warning */
2509 dyninf.dyn_rel_off = 0; /* avoid warning */
2510
2511 if (file_type != TCC_OUTPUT_OBJ) {
2512 relocate_common_syms();
2513
2514 tcc_add_linker_symbols(s1);
2515
2516 if (!s1->static_link) {
2517 if (file_type == TCC_OUTPUT_EXE) {
2518 char *ptr;
2519 /* allow override the dynamic loader */
2520 const char *elfint = getenv("LD_SO");
2521 if (elfint == NULL)
2522 elfint = DEFAULT_ELFINTERP(s1);
2523 /* add interpreter section only if executable */
2524 interp = new_section(s1, ".interp", SHT_PROGBITS, SHF_ALLOC);
2525 interp->sh_addralign = 1;
2526 ptr = section_ptr_add(interp, 1 + strlen(elfint));
2527 strcpy(ptr, elfint);
2528 }
2529
2530 /* add dynamic symbol table */
2531 s1->dynsym = new_symtab(s1, ".dynsym", SHT_DYNSYM, SHF_ALLOC,
2532 ".dynstr",
2533 ".hash", SHF_ALLOC);
2534 dynstr = s1->dynsym->link;
2535
2536 /* add dynamic section */
2537 dynamic = new_section(s1, ".dynamic", SHT_DYNAMIC,
2538 SHF_ALLOC | SHF_WRITE);
2539 dynamic->link = dynstr;
2540 dynamic->sh_entsize = sizeof(ElfW(Dyn));
2541
2542 build_got(s1);
2543
2544 if (file_type == TCC_OUTPUT_EXE) {
2545 bind_exe_dynsyms(s1);
2546
2547 if (s1->nb_errors) {
2548 ret = -1;
2549 goto the_end;
2550 }
2551
2552 bind_libs_dynsyms(s1);
2553 } else /* shared library case: simply export all global symbols */
2554 export_global_syms(s1);
2555
2556 build_got_entries(s1);
2557
2558 /* add a list of needed dlls */
2559 for(i = 0; i < s1->nb_loaded_dlls; i++) {
2560 DLLReference *dllref = s1->loaded_dlls[i];
2561 if (dllref->level == 0)
2562 put_dt(dynamic, DT_NEEDED, put_elf_str(dynstr, dllref->name));
2563 }
2564
2565 if (s1->rpath)
2566 put_dt(dynamic, DT_RPATH, put_elf_str(dynstr, s1->rpath));
2567
2568 /* XXX: currently, since we do not handle PIC code, we
2569 must relocate the readonly segments */
2570 if (file_type == TCC_OUTPUT_DLL) {
2571 if (s1->soname)
2572 put_dt(dynamic, DT_SONAME, put_elf_str(dynstr, s1->soname));
2573 put_dt(dynamic, DT_TEXTREL, 0);
2574 }
2575
2576 if (s1->symbolic)
2577 put_dt(dynamic, DT_SYMBOLIC, 0);
2578
2579 /* add necessary space for other entries */
2580 dyninf.dyn_rel_off = dynamic->data_offset;
2581 dynamic->data_offset += sizeof(ElfW(Dyn)) * EXTRA_RELITEMS;
2582 } else {
2583 /* still need to build got entries in case of static link */
2584 build_got_entries(s1);
2585 }
2586 }
2587
2588 /* we add a section for symbols */
2589 strsec = new_section(s1, ".shstrtab", SHT_STRTAB, 0);
2590 put_elf_str(strsec, "");
2591
2592 /* compute number of sections */
2593 shnum = s1->nb_sections;
2594
2595 /* this array is used to reorder sections in the output file */
2596 sec_order = tcc_malloc(sizeof(int) * shnum);
2597 sec_order[0] = 0;
2598
2599 /* compute number of program headers */
2600 switch(file_type) {
2601 default:
2602 case TCC_OUTPUT_OBJ:
2603 phnum = 0;
2604 break;
2605 case TCC_OUTPUT_EXE:
2606 if (!s1->static_link)
2607 phnum = 4 + HAVE_PHDR;
2608 else
2609 phnum = 2;
2610 break;
2611 case TCC_OUTPUT_DLL:
2612 phnum = 3;
2613 break;
2614 }
2615
2616 /* Allocate strings for section names */
2617 alloc_sec_names(s1, file_type, strsec);
2618
2619 /* allocate program segment headers */
2620 phdr = tcc_mallocz(phnum * sizeof(ElfW(Phdr)));
2621
2622 /* compute section to program header mapping */
2623 file_offset = layout_sections(s1, phdr, phnum, interp, strsec, &dyninf,
2624 sec_order);
2625
2626 /* Fill remaining program header and finalize relocation related to dynamic
2627 linking. */
2628 if (phnum > 0) {
2629 fill_unloadable_phdr(phdr, phnum, interp, dynamic);
2630 if (dynamic) {
2631 dyninf.dynamic = dynamic;
2632 dyninf.dynstr = dynstr;
2633
2634 fill_dynamic(s1, &dyninf);
2635
2636 /* put in GOT the dynamic section address and relocate PLT */
2637 write32le(s1->got->data, dynamic->sh_addr);
2638 if (file_type == TCC_OUTPUT_EXE
2639 #if defined(TCC_OUTPUT_DLL_WITH_PLT)
2640 || file_type == TCC_OUTPUT_DLL
2641 #endif
2642 )
2643 relocate_plt(s1);
2644
2645 /* relocate symbols in .dynsym now that final addresses are known */
2646 for_each_elem(s1->dynsym, 1, sym, ElfW(Sym)) {
2647 if (sym->st_shndx == SHN_UNDEF) {
2648 /* relocate to PLT if symbol corresponds to a PLT entry,
2649 but not if it's a weak symbol */
2650 if (ELFW(ST_BIND)(sym->st_info) == STB_WEAK)
2651 sym->st_value = 0;
2652 else if (sym->st_value)
2653 sym->st_value += s1->plt->sh_addr;
2654 } else if (sym->st_shndx < SHN_LORESERVE) {
2655 /* do symbol relocation */
2656 sym->st_value += s1->sections[sym->st_shndx]->sh_addr;
2657 }
2658 }
2659 }
2660 }
2661
2662 /* if building executable or DLL, then relocate each section
2663 except the GOT which is already relocated */
2664 if (file_type != TCC_OUTPUT_OBJ) {
2665 ret = final_sections_reloc(s1);
2666 if (ret)
2667 goto the_end;
2668 }
2669
2670 /* Perform relocation to GOT or PLT entries */
2671 if (file_type == TCC_OUTPUT_EXE && s1->static_link)
2672 fill_got(s1);
2673
2674 /* Create the ELF file with name 'filename' */
2675 ret = tcc_write_elf_file(s1, filename, phnum, phdr, file_offset, sec_order);
2676 the_end:
2677 tcc_free(s1->symtab_to_dynsym);
2678 tcc_free(sec_order);
2679 tcc_free(phdr);
2680 tcc_free(s1->sym_attrs);
2681 s1->sym_attrs = NULL;
2682 return ret;
2683 }
2684
2685 LIBTCCAPI int tcc_output_file(TCCState *s, const char *filename)
2686 {
2687 int ret;
2688 #ifdef TCC_TARGET_PE
2689 if (s->output_type != TCC_OUTPUT_OBJ) {
2690 ret = pe_output_file(s, filename);
2691 } else
2692 #endif
2693 ret = elf_output_file(s, filename);
2694 return ret;
2695 }
2696
2697 static void *load_data(int fd, unsigned long file_offset, unsigned long size)
2698 {
2699 void *data;
2700
2701 data = tcc_malloc(size);
2702 lseek(fd, file_offset, SEEK_SET);
2703 read(fd, data, size);
2704 return data;
2705 }
2706
2707 typedef struct SectionMergeInfo {
2708 Section *s; /* corresponding existing section */
2709 unsigned long offset; /* offset of the new section in the existing section */
2710 uint8_t new_section; /* true if section 's' was added */
2711 uint8_t link_once; /* true if link once section */
2712 } SectionMergeInfo;
2713
2714 ST_FUNC int tcc_object_type(int fd, ElfW(Ehdr) *h)
2715 {
2716 int size = read(fd, h, sizeof *h);
2717 if (size == sizeof *h && 0 == memcmp(h, ELFMAG, 4)) {
2718 if (h->e_type == ET_REL)
2719 return AFF_BINTYPE_REL;
2720 if (h->e_type == ET_DYN)
2721 return AFF_BINTYPE_DYN;
2722 } else if (size >= 8) {
2723 if (0 == memcmp(h, ARMAG, 8))
2724 return AFF_BINTYPE_AR;
2725 #ifdef TCC_TARGET_COFF
2726 if (((struct filehdr*)h)->f_magic == COFF_C67_MAGIC)
2727 return AFF_BINTYPE_C67;
2728 #endif
2729 }
2730 return 0;
2731 }
2732
2733 /* load an object file and merge it with current files */
2734 /* XXX: handle correctly stab (debug) info */
2735 ST_FUNC int tcc_load_object_file(TCCState *s1,
2736 int fd, unsigned long file_offset)
2737 {
2738 ElfW(Ehdr) ehdr;
2739 ElfW(Shdr) *shdr, *sh;
2740 int size, i, j, offset, offseti, nb_syms, sym_index, ret;
2741 unsigned char *strsec, *strtab;
2742 int *old_to_new_syms;
2743 char *sh_name, *name;
2744 SectionMergeInfo *sm_table, *sm;
2745 ElfW(Sym) *sym, *symtab;
2746 ElfW_Rel *rel;
2747 Section *s;
2748
2749 int stab_index;
2750 int stabstr_index;
2751
2752 stab_index = stabstr_index = 0;
2753
2754 lseek(fd, file_offset, SEEK_SET);
2755 if (tcc_object_type(fd, &ehdr) != AFF_BINTYPE_REL)
2756 goto fail1;
2757 /* test CPU specific stuff */
2758 if (ehdr.e_ident[5] != ELFDATA2LSB ||
2759 ehdr.e_machine != EM_TCC_TARGET) {
2760 fail1:
2761 tcc_error_noabort("invalid object file");
2762 return -1;
2763 }
2764 /* read sections */
2765 shdr = load_data(fd, file_offset + ehdr.e_shoff,
2766 sizeof(ElfW(Shdr)) * ehdr.e_shnum);
2767 sm_table = tcc_mallocz(sizeof(SectionMergeInfo) * ehdr.e_shnum);
2768
2769 /* load section names */
2770 sh = &shdr[ehdr.e_shstrndx];
2771 strsec = load_data(fd, file_offset + sh->sh_offset, sh->sh_size);
2772
2773 /* load symtab and strtab */
2774 old_to_new_syms = NULL;
2775 symtab = NULL;
2776 strtab = NULL;
2777 nb_syms = 0;
2778 for(i = 1; i < ehdr.e_shnum; i++) {
2779 sh = &shdr[i];
2780 if (sh->sh_type == SHT_SYMTAB) {
2781 if (symtab) {
2782 tcc_error_noabort("object must contain only one symtab");
2783 fail:
2784 ret = -1;
2785 goto the_end;
2786 }
2787 nb_syms = sh->sh_size / sizeof(ElfW(Sym));
2788 symtab = load_data(fd, file_offset + sh->sh_offset, sh->sh_size);
2789 sm_table[i].s = symtab_section;
2790
2791 /* now load strtab */
2792 sh = &shdr[sh->sh_link];
2793 strtab = load_data(fd, file_offset + sh->sh_offset, sh->sh_size);
2794 }
2795 }
2796
2797 /* now examine each section and try to merge its content with the
2798 ones in memory */
2799 for(i = 1; i < ehdr.e_shnum; i++) {
2800 /* no need to examine section name strtab */
2801 if (i == ehdr.e_shstrndx)
2802 continue;
2803 sh = &shdr[i];
2804 sh_name = (char *) strsec + sh->sh_name;
2805 /* ignore sections types we do not handle */
2806 if (sh->sh_type != SHT_PROGBITS &&
2807 sh->sh_type != SHT_RELX &&
2808 #ifdef TCC_ARM_EABI
2809 sh->sh_type != SHT_ARM_EXIDX &&
2810 #endif
2811 sh->sh_type != SHT_NOBITS &&
2812 sh->sh_type != SHT_PREINIT_ARRAY &&
2813 sh->sh_type != SHT_INIT_ARRAY &&
2814 sh->sh_type != SHT_FINI_ARRAY &&
2815 strcmp(sh_name, ".stabstr")
2816 )
2817 continue;
2818 if (sh->sh_addralign < 1)
2819 sh->sh_addralign = 1;
2820 /* find corresponding section, if any */
2821 for(j = 1; j < s1->nb_sections;j++) {
2822 s = s1->sections[j];
2823 if (!strcmp(s->name, sh_name)) {
2824 if (!strncmp(sh_name, ".gnu.linkonce",
2825 sizeof(".gnu.linkonce") - 1)) {
2826 /* if a 'linkonce' section is already present, we
2827 do not add it again. It is a little tricky as
2828 symbols can still be defined in
2829 it. */
2830 sm_table[i].link_once = 1;
2831 goto next;
2832 } else {
2833 goto found;
2834 }
2835 }
2836 }
2837 /* not found: create new section */
2838 s = new_section(s1, sh_name, sh->sh_type, sh->sh_flags);
2839 /* take as much info as possible from the section. sh_link and
2840 sh_info will be updated later */
2841 s->sh_addralign = sh->sh_addralign;
2842 s->sh_entsize = sh->sh_entsize;
2843 sm_table[i].new_section = 1;
2844 found:
2845 if (sh->sh_type != s->sh_type) {
2846 tcc_error_noabort("invalid section type");
2847 goto fail;
2848 }
2849
2850 /* align start of section */
2851 offset = s->data_offset;
2852
2853 if (0 == strcmp(sh_name, ".stab")) {
2854 stab_index = i;
2855 goto no_align;
2856 }
2857 if (0 == strcmp(sh_name, ".stabstr")) {
2858 stabstr_index = i;
2859 goto no_align;
2860 }
2861
2862 size = sh->sh_addralign - 1;
2863 offset = (offset + size) & ~size;
2864 if (sh->sh_addralign > s->sh_addralign)
2865 s->sh_addralign = sh->sh_addralign;
2866 s->data_offset = offset;
2867 no_align:
2868 sm_table[i].offset = offset;
2869 sm_table[i].s = s;
2870 /* concatenate sections */
2871 size = sh->sh_size;
2872 if (sh->sh_type != SHT_NOBITS) {
2873 unsigned char *ptr;
2874 lseek(fd, file_offset + sh->sh_offset, SEEK_SET);
2875 ptr = section_ptr_add(s, size);
2876 read(fd, ptr, size);
2877 } else {
2878 s->data_offset += size;
2879 }
2880 next: ;
2881 }
2882
2883 /* gr relocate stab strings */
2884 if (stab_index && stabstr_index) {
2885 Stab_Sym *a, *b;
2886 unsigned o;
2887 s = sm_table[stab_index].s;
2888 a = (Stab_Sym *)(s->data + sm_table[stab_index].offset);
2889 b = (Stab_Sym *)(s->data + s->data_offset);
2890 o = sm_table[stabstr_index].offset;
2891 while (a < b)
2892 a->n_strx += o, a++;
2893 }
2894
2895 /* second short pass to update sh_link and sh_info fields of new
2896 sections */
2897 for(i = 1; i < ehdr.e_shnum; i++) {
2898 s = sm_table[i].s;
2899 if (!s || !sm_table[i].new_section)
2900 continue;
2901 sh = &shdr[i];
2902 if (sh->sh_link > 0)
2903 s->link = sm_table[sh->sh_link].s;
2904 if (sh->sh_type == SHT_RELX) {
2905 s->sh_info = sm_table[sh->sh_info].s->sh_num;
2906 /* update backward link */
2907 s1->sections[s->sh_info]->reloc = s;
2908 }
2909 }
2910 sm = sm_table;
2911
2912 /* resolve symbols */
2913 old_to_new_syms = tcc_mallocz(nb_syms * sizeof(int));
2914
2915 sym = symtab + 1;
2916 for(i = 1; i < nb_syms; i++, sym++) {
2917 if (sym->st_shndx != SHN_UNDEF &&
2918 sym->st_shndx < SHN_LORESERVE) {
2919 sm = &sm_table[sym->st_shndx];
2920 if (sm->link_once) {
2921 /* if a symbol is in a link once section, we use the
2922 already defined symbol. It is very important to get
2923 correct relocations */
2924 if (ELFW(ST_BIND)(sym->st_info) != STB_LOCAL) {
2925 name = (char *) strtab + sym->st_name;
2926 sym_index = find_elf_sym(symtab_section, name);
2927 if (sym_index)
2928 old_to_new_syms[i] = sym_index;
2929 }
2930 continue;
2931 }
2932 /* if no corresponding section added, no need to add symbol */
2933 if (!sm->s)
2934 continue;
2935 /* convert section number */
2936 sym->st_shndx = sm->s->sh_num;
2937 /* offset value */
2938 sym->st_value += sm->offset;
2939 }
2940 /* add symbol */
2941 name = (char *) strtab + sym->st_name;
2942 sym_index = add_elf_sym(symtab_section, sym->st_value, sym->st_size,
2943 sym->st_info, sym->st_other,
2944 sym->st_shndx, name);
2945 old_to_new_syms[i] = sym_index;
2946 }
2947
2948 /* third pass to patch relocation entries */
2949 for(i = 1; i < ehdr.e_shnum; i++) {
2950 s = sm_table[i].s;
2951 if (!s)
2952 continue;
2953 sh = &shdr[i];
2954 offset = sm_table[i].offset;
2955 switch(s->sh_type) {
2956 case SHT_RELX:
2957 /* take relocation offset information */
2958 offseti = sm_table[sh->sh_info].offset;
2959 for_each_elem(s, (offset / sizeof(*rel)), rel, ElfW_Rel) {
2960 int type;
2961 unsigned sym_index;
2962 /* convert symbol index */
2963 type = ELFW(R_TYPE)(rel->r_info);
2964 sym_index = ELFW(R_SYM)(rel->r_info);
2965 /* NOTE: only one symtab assumed */
2966 if (sym_index >= nb_syms)
2967 goto invalid_reloc;
2968 sym_index = old_to_new_syms[sym_index];
2969 /* ignore link_once in rel section. */
2970 if (!sym_index && !sm->link_once
2971 #ifdef TCC_TARGET_ARM
2972 && type != R_ARM_V4BX
2973 #endif
2974 ) {
2975 invalid_reloc:
2976 tcc_error_noabort("Invalid relocation entry [%2d] '%s' @ %.8x",
2977 i, strsec + sh->sh_name, rel->r_offset);
2978 goto fail;
2979 }
2980 rel->r_info = ELFW(R_INFO)(sym_index, type);
2981 /* offset the relocation offset */
2982 rel->r_offset += offseti;
2983 #ifdef TCC_TARGET_ARM
2984 /* Jumps and branches from a Thumb code to a PLT entry need
2985 special handling since PLT entries are ARM code.
2986 Unconditional bl instructions referencing PLT entries are
2987 handled by converting these instructions into blx
2988 instructions. Other case of instructions referencing a PLT
2989 entry require to add a Thumb stub before the PLT entry to
2990 switch to ARM mode. We set bit plt_thumb_stub of the
2991 attribute of a symbol to indicate such a case. */
2992 if (type == R_ARM_THM_JUMP24)
2993 alloc_sym_attr(s1, sym_index)->plt_thumb_stub = 1;
2994 #endif
2995 }
2996 break;
2997 default:
2998 break;
2999 }
3000 }
3001
3002 ret = 0;
3003 the_end:
3004 tcc_free(symtab);
3005 tcc_free(strtab);
3006 tcc_free(old_to_new_syms);
3007 tcc_free(sm_table);
3008 tcc_free(strsec);
3009 tcc_free(shdr);
3010 return ret;
3011 }
3012
3013 typedef struct ArchiveHeader {
3014 char ar_name[16]; /* name of this member */
3015 char ar_date[12]; /* file mtime */
3016 char ar_uid[6]; /* owner uid; printed as decimal */
3017 char ar_gid[6]; /* owner gid; printed as decimal */
3018 char ar_mode[8]; /* file mode, printed as octal */
3019 char ar_size[10]; /* file size, printed as decimal */
3020 char ar_fmag[2]; /* should contain ARFMAG */
3021 } ArchiveHeader;
3022
3023 static int get_be32(const uint8_t *b)
3024 {
3025 return b[3] | (b[2] << 8) | (b[1] << 16) | (b[0] << 24);
3026 }
3027
3028 /* load only the objects which resolve undefined symbols */
3029 static int tcc_load_alacarte(TCCState *s1, int fd, int size)
3030 {
3031 int i, bound, nsyms, sym_index, off, ret;
3032 uint8_t *data;
3033 const char *ar_names, *p;
3034 const uint8_t *ar_index;
3035 ElfW(Sym) *sym;
3036
3037 data = tcc_malloc(size);
3038 if (read(fd, data, size) != size)
3039 goto fail;
3040 nsyms = get_be32(data);
3041 ar_index = data + 4;
3042 ar_names = (char *) ar_index + nsyms * 4;
3043
3044 do {
3045 bound = 0;
3046 for(p = ar_names, i = 0; i < nsyms; i++, p += strlen(p)+1) {
3047 sym_index = find_elf_sym(symtab_section, p);
3048 if(sym_index) {
3049 sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
3050 if(sym->st_shndx == SHN_UNDEF) {
3051 off = get_be32(ar_index + i * 4) + sizeof(ArchiveHeader);
3052 ++bound;
3053 if(tcc_load_object_file(s1, fd, off) < 0) {
3054 fail:
3055 ret = -1;
3056 goto the_end;
3057 }
3058 }
3059 }
3060 }
3061 } while(bound);
3062 ret = 0;
3063 the_end:
3064 tcc_free(data);
3065 return ret;
3066 }
3067
3068 /* load a '.a' file */
3069 ST_FUNC int tcc_load_archive(TCCState *s1, int fd)
3070 {
3071 ArchiveHeader hdr;
3072 char ar_size[11];
3073 char ar_name[17];
3074 char magic[8];
3075 int size, len, i;
3076 unsigned long file_offset;
3077
3078 /* skip magic which was already checked */
3079 read(fd, magic, sizeof(magic));
3080
3081 for(;;) {
3082 len = read(fd, &hdr, sizeof(hdr));
3083 if (len == 0)
3084 break;
3085 if (len != sizeof(hdr)) {
3086 tcc_error_noabort("invalid archive");
3087 return -1;
3088 }
3089 memcpy(ar_size, hdr.ar_size, sizeof(hdr.ar_size));
3090 ar_size[sizeof(hdr.ar_size)] = '\0';
3091 size = strtol(ar_size, NULL, 0);
3092 memcpy(ar_name, hdr.ar_name, sizeof(hdr.ar_name));
3093 for(i = sizeof(hdr.ar_name) - 1; i >= 0; i--) {
3094 if (ar_name[i] != ' ')
3095 break;
3096 }
3097 ar_name[i + 1] = '\0';
3098 file_offset = lseek(fd, 0, SEEK_CUR);
3099 /* align to even */
3100 size = (size + 1) & ~1;
3101 if (!strcmp(ar_name, "/")) {
3102 /* coff symbol table : we handle it */
3103 if(s1->alacarte_link)
3104 return tcc_load_alacarte(s1, fd, size);
3105 } else {
3106 ElfW(Ehdr) ehdr;
3107 if (tcc_object_type(fd, &ehdr) == AFF_BINTYPE_REL) {
3108 if (tcc_load_object_file(s1, fd, file_offset) < 0)
3109 return -1;
3110 }
3111 }
3112 lseek(fd, file_offset + size, SEEK_SET);
3113 }
3114 return 0;
3115 }
3116
3117 #ifndef TCC_TARGET_PE
3118 /* load a DLL and all referenced DLLs. 'level = 0' means that the DLL
3119 is referenced by the user (so it should be added as DT_NEEDED in
3120 the generated ELF file) */
3121 ST_FUNC int tcc_load_dll(TCCState *s1, int fd, const char *filename, int level)
3122 {
3123 ElfW(Ehdr) ehdr;
3124 ElfW(Shdr) *shdr, *sh, *sh1;
3125 int i, j, nb_syms, nb_dts, sym_bind, ret;
3126 ElfW(Sym) *sym, *dynsym;
3127 ElfW(Dyn) *dt, *dynamic;
3128 unsigned char *dynstr;
3129 const char *name, *soname;
3130 DLLReference *dllref;
3131
3132 read(fd, &ehdr, sizeof(ehdr));
3133
3134 /* test CPU specific stuff */
3135 if (ehdr.e_ident[5] != ELFDATA2LSB ||
3136 ehdr.e_machine != EM_TCC_TARGET) {
3137 tcc_error_noabort("bad architecture");
3138 return -1;
3139 }
3140
3141 /* read sections */
3142 shdr = load_data(fd, ehdr.e_shoff, sizeof(ElfW(Shdr)) * ehdr.e_shnum);
3143
3144 /* load dynamic section and dynamic symbols */
3145 nb_syms = 0;
3146 nb_dts = 0;
3147 dynamic = NULL;
3148 dynsym = NULL; /* avoid warning */
3149 dynstr = NULL; /* avoid warning */
3150 for(i = 0, sh = shdr; i < ehdr.e_shnum; i++, sh++) {
3151 switch(sh->sh_type) {
3152 case SHT_DYNAMIC:
3153 nb_dts = sh->sh_size / sizeof(ElfW(Dyn));
3154 dynamic = load_data(fd, sh->sh_offset, sh->sh_size);
3155 break;
3156 case SHT_DYNSYM:
3157 nb_syms = sh->sh_size / sizeof(ElfW(Sym));
3158 dynsym = load_data(fd, sh->sh_offset, sh->sh_size);
3159 sh1 = &shdr[sh->sh_link];
3160 dynstr = load_data(fd, sh1->sh_offset, sh1->sh_size);
3161 break;
3162 default:
3163 break;
3164 }
3165 }
3166
3167 /* compute the real library name */
3168 soname = tcc_basename(filename);
3169
3170 for(i = 0, dt = dynamic; i < nb_dts; i++, dt++) {
3171 if (dt->d_tag == DT_SONAME) {
3172 soname = (char *) dynstr + dt->d_un.d_val;
3173 }
3174 }
3175
3176 /* if the dll is already loaded, do not load it */
3177 for(i = 0; i < s1->nb_loaded_dlls; i++) {
3178 dllref = s1->loaded_dlls[i];
3179 if (!strcmp(soname, dllref->name)) {
3180 /* but update level if needed */
3181 if (level < dllref->level)
3182 dllref->level = level;
3183 ret = 0;
3184 goto the_end;
3185 }
3186 }
3187
3188 /* add the dll and its level */
3189 dllref = tcc_mallocz(sizeof(DLLReference) + strlen(soname));
3190 dllref->level = level;
3191 strcpy(dllref->name, soname);
3192 dynarray_add((void ***)&s1->loaded_dlls, &s1->nb_loaded_dlls, dllref);
3193
3194 /* add dynamic symbols in dynsym_section */
3195 for(i = 1, sym = dynsym + 1; i < nb_syms; i++, sym++) {
3196 sym_bind = ELFW(ST_BIND)(sym->st_info);
3197 if (sym_bind == STB_LOCAL)
3198 continue;
3199 name = (char *) dynstr + sym->st_name;
3200 add_elf_sym(s1->dynsymtab_section, sym->st_value, sym->st_size,
3201 sym->st_info, sym->st_other, sym->st_shndx, name);
3202 }
3203
3204 /* load all referenced DLLs */
3205 for(i = 0, dt = dynamic; i < nb_dts; i++, dt++) {
3206 switch(dt->d_tag) {
3207 case DT_NEEDED:
3208 name = (char *) dynstr + dt->d_un.d_val;
3209 for(j = 0; j < s1->nb_loaded_dlls; j++) {
3210 dllref = s1->loaded_dlls[j];
3211 if (!strcmp(name, dllref->name))
3212 goto already_loaded;
3213 }
3214 if (tcc_add_dll(s1, name, AFF_REFERENCED_DLL) < 0) {
3215 tcc_error_noabort("referenced dll '%s' not found", name);
3216 ret = -1;
3217 goto the_end;
3218 }
3219 already_loaded:
3220 break;
3221 }
3222 }
3223 ret = 0;
3224 the_end:
3225 tcc_free(dynstr);
3226 tcc_free(dynsym);
3227 tcc_free(dynamic);
3228 tcc_free(shdr);
3229 return ret;
3230 }
3231
3232 #define LD_TOK_NAME 256
3233 #define LD_TOK_EOF (-1)
3234
3235 /* return next ld script token */
3236 static int ld_next(TCCState *s1, char *name, int name_size)
3237 {
3238 int c;
3239 char *q;
3240
3241 redo:
3242 switch(ch) {
3243 case ' ':
3244 case '\t':
3245 case '\f':
3246 case '\v':
3247 case '\r':
3248 case '\n':
3249 inp();
3250 goto redo;
3251 case '/':
3252 minp();
3253 if (ch == '*') {
3254 file->buf_ptr = parse_comment(file->buf_ptr);
3255 ch = file->buf_ptr[0];
3256 goto redo;
3257 } else {
3258 q = name;
3259 *q++ = '/';
3260 goto parse_name;
3261 }
3262 break;
3263 case '\\':
3264 ch = handle_eob();
3265 if (ch != '\\')
3266 goto redo;
3267 /* fall through */
3268 /* case 'a' ... 'z': */
3269 case 'a':
3270 case 'b':
3271 case 'c':
3272 case 'd':
3273 case 'e':
3274 case 'f':
3275 case 'g':
3276 case 'h':
3277 case 'i':
3278 case 'j':
3279 case 'k':
3280 case 'l':
3281 case 'm':
3282 case 'n':
3283 case 'o':
3284 case 'p':
3285 case 'q':
3286 case 'r':
3287 case 's':
3288 case 't':
3289 case 'u':
3290 case 'v':
3291 case 'w':
3292 case 'x':
3293 case 'y':
3294 case 'z':
3295 /* case 'A' ... 'z': */
3296 case 'A':
3297 case 'B':
3298 case 'C':
3299 case 'D':
3300 case 'E':
3301 case 'F':
3302 case 'G':
3303 case 'H':
3304 case 'I':
3305 case 'J':
3306 case 'K':
3307 case 'L':
3308 case 'M':
3309 case 'N':
3310 case 'O':
3311 case 'P':
3312 case 'Q':
3313 case 'R':
3314 case 'S':
3315 case 'T':
3316 case 'U':
3317 case 'V':
3318 case 'W':
3319 case 'X':
3320 case 'Y':
3321 case 'Z':
3322 case '_':
3323 case '.':
3324 case '$':
3325 case '~':
3326 q = name;
3327 parse_name:
3328 for(;;) {
3329 if (!((ch >= 'a' && ch <= 'z') ||
3330 (ch >= 'A' && ch <= 'Z') ||
3331 (ch >= '0' && ch <= '9') ||
3332 strchr("/.-_+=$:\\,~", ch)))
3333 break;
3334 if ((q - name) < name_size - 1) {
3335 *q++ = ch;
3336 }
3337 minp();
3338 }
3339 *q = '\0';
3340 c = LD_TOK_NAME;
3341 break;
3342 case CH_EOF:
3343 c = LD_TOK_EOF;
3344 break;
3345 default:
3346 c = ch;
3347 inp();
3348 break;
3349 }
3350 return c;
3351 }
3352
3353 static int ld_add_file(TCCState *s1, const char filename[])
3354 {
3355 int ret;
3356
3357 ret = tcc_add_file_internal(s1, filename, AFF_TYPE_BIN);
3358 if (ret)
3359 ret = tcc_add_dll(s1, filename, 0);
3360 return ret;
3361 }
3362
3363 static inline int new_undef_syms(void)
3364 {
3365 int ret = 0;
3366 ret = new_undef_sym;
3367 new_undef_sym = 0;
3368 return ret;
3369 }
3370
3371 static int ld_add_file_list(TCCState *s1, const char *cmd, int as_needed)
3372 {
3373 char filename[1024], libname[1024];
3374 int t, group, nblibs = 0, ret = 0;
3375 char **libs = NULL;
3376
3377 group = !strcmp(cmd, "GROUP");
3378 if (!as_needed)
3379 new_undef_syms();
3380 t = ld_next(s1, filename, sizeof(filename));
3381 if (t != '(')
3382 expect("(");
3383 t = ld_next(s1, filename, sizeof(filename));
3384 for(;;) {
3385 libname[0] = '\0';
3386 if (t == LD_TOK_EOF) {
3387 tcc_error_noabort("unexpected end of file");
3388 ret = -1;
3389 goto lib_parse_error;
3390 } else if (t == ')') {
3391 break;
3392 } else if (t == '-') {
3393 t = ld_next(s1, filename, sizeof(filename));
3394 if ((t != LD_TOK_NAME) || (filename[0] != 'l')) {
3395 tcc_error_noabort("library name expected");
3396 ret = -1;
3397 goto lib_parse_error;
3398 }
3399 pstrcpy(libname, sizeof libname, &filename[1]);
3400 if (s1->static_link) {
3401 snprintf(filename, sizeof filename, "lib%s.a", libname);
3402 } else {
3403 snprintf(filename, sizeof filename, "lib%s.so", libname);
3404 }
3405 } else if (t != LD_TOK_NAME) {
3406 tcc_error_noabort("filename expected");
3407 ret = -1;
3408 goto lib_parse_error;
3409 }
3410 if (!strcmp(filename, "AS_NEEDED")) {
3411 ret = ld_add_file_list(s1, cmd, 1);
3412 if (ret)
3413 goto lib_parse_error;
3414 } else {
3415 /* TODO: Implement AS_NEEDED support. Ignore it for now */
3416 if (!as_needed) {
3417 ret = ld_add_file(s1, filename);
3418 if (ret)
3419 goto lib_parse_error;
3420 if (group) {
3421 /* Add the filename *and* the libname to avoid future conversions */
3422 dynarray_add((void ***) &libs, &nblibs, tcc_strdup(filename));
3423 if (libname[0] != '\0')
3424 dynarray_add((void ***) &libs, &nblibs, tcc_strdup(libname));
3425 }
3426 }
3427 }
3428 t = ld_next(s1, filename, sizeof(filename));
3429 if (t == ',') {
3430 t = ld_next(s1, filename, sizeof(filename));
3431 }
3432 }
3433 if (group && !as_needed) {
3434 while (new_undef_syms()) {
3435 int i;
3436
3437 for (i = 0; i < nblibs; i ++)
3438 ld_add_file(s1, libs[i]);
3439 }
3440 }
3441 lib_parse_error:
3442 dynarray_reset(&libs, &nblibs);
3443 return ret;
3444 }
3445
3446 /* interpret a subset of GNU ldscripts to handle the dummy libc.so
3447 files */
3448 ST_FUNC int tcc_load_ldscript(TCCState *s1)
3449 {
3450 char cmd[64];
3451 char filename[1024];
3452 int t, ret;
3453
3454 ch = handle_eob();
3455 for(;;) {
3456 t = ld_next(s1, cmd, sizeof(cmd));
3457 if (t == LD_TOK_EOF)
3458 return 0;
3459 else if (t != LD_TOK_NAME)
3460 return -1;
3461 if (!strcmp(cmd, "INPUT") ||
3462 !strcmp(cmd, "GROUP")) {
3463 ret = ld_add_file_list(s1, cmd, 0);
3464 if (ret)
3465 return ret;
3466 } else if (!strcmp(cmd, "OUTPUT_FORMAT") ||
3467 !strcmp(cmd, "TARGET")) {
3468 /* ignore some commands */
3469 t = ld_next(s1, cmd, sizeof(cmd));
3470 if (t != '(')
3471 expect("(");
3472 for(;;) {
3473 t = ld_next(s1, filename, sizeof(filename));
3474 if (t == LD_TOK_EOF) {
3475 tcc_error_noabort("unexpected end of file");
3476 return -1;
3477 } else if (t == ')') {
3478 break;
3479 }
3480 }
3481 } else {
3482 return -1;
3483 }
3484 }
3485 return 0;
3486 }
3487 #endif /* !TCC_TARGET_PE */
Tiny C Compiler main repository